Biology of Human Sodium Glucose Transporters

advertisement
Physiol Rev 91: 733–794, 2011;
doi:10.1152/physrev.00055.2009.
Biology of Human Sodium Glucose Transporters
ERNEST M. WRIGHT, DONALD D. F. LOO, AND BRUCE A. HIRAYAMA
Department of Physiology, David Geffen School of Medicine at University of California Los Angeles,
Los Angeles, California
I. Introduction
II. 1987 BC
III. Expression Cloning
A. The intestinal brush-border transporter
B. Stoichiometry of Na⫹ to glucose transport
C. Reversible transport
IV. Genes
A. Chromosomal location
B. Gene mapping
C. Human gene family (SLC5)
D. Functional characterization
V. Expression
A. mRNA
B. Proteins
C. PET imaging
VI. Prokaryote SGLTs
VII. Structure
A. Amino acid sequences
B. Secondary structure
C. Monomer
D. Protein
E. Circular dichroism
F. Crystal structure
G. LeuT superfamily
VIII. Sugar Selectivity
A. Monosaccharides
B. Glucosides
C. Inhibitors
IX. Ion Selectivity
X. Kinetics
A. Steady-state kinetics
B. Pre-steady-state kinetics
C. Conformational dynamics
D. Substrate and drug interactions
XI. Kinetic Modeling
A. Model
B. Experimental basis
C. Voltage sensitivity
D. Estimating parameters
E. Testing
F. Distribution of conformations
XII. Structure and Function
XIII. Multifunctional Proteins
A. Na⫹ uniport
B. Water and urea channels
C. Coupled water, urea, and glucose transport
D. Glucose sensor
E. Glucose sensing in the gut
F. Glucose sensing in the brain
XIV. Physiology and Pathophysiology
www.prv.org
734
735
737
737
738
739
741
741
741
741
742
743
743
744
745
746
747
747
748
749
749
749
750
751
752
752
754
755
757
758
758
761
767
768
769
769
769
770
770
770
771
773
774
774
775
775
775
776
776
777
733
734
WRIGHT, LOO, AND HIRAYAMA
A. Regulation of expression
B. Intestinal absorption
C. Oral rehydration therapy
D. Glucose galactose malabsorption
E. Enteric infection
F. Renal reabsorption
G. Familial renal glucosuria
H. Bile, milk, and saliva
I. Cancer
J. Diabetes
XV. Outlook
XVI. Unresolved Problems
777
778
780
780
782
782
783
785
785
786
787
787
Wright EM, Loo DDF, Hirayama BA. Biology of Human Sodium Glucose Transporters. Physiol Rev 91: 733–794,
2011; doi:10.1152/physrev.00055.2009.—There are two classes of glucose transporters involved in glucose homeostasis in the body, the facilitated transporters or uniporters (GLUTs) and the active transporters or symporters
(SGLTs). The energy for active glucose transport is provided by the sodium gradient across the cell membrane, the
Na⫹ glucose cotransport hypothesis first proposed in 1960 by Crane. Since the cloning of SGLT1 in 1987, there have
been advances in the genetics, molecular biology, biochemistry, biophysics, and structure of SGLTs. There are 12
members of the human SGLT (SLC5) gene family, including cotransporters for sugars, anions, vitamins, and
short-chain fatty acids. Here we give a personal review of these advances. The SGLTs belong to a structural class
of membrane proteins from unrelated gene families of antiporters and Na⫹ and H⫹ symporters. This class shares a
common atomic architecture and a common transport mechanism. SGLTs also function as water and urea channels,
glucose sensors, and coupled-water and urea transporters. We also discuss the physiology and pathophysiology of
SGLTs, e.g., glucose galactose malabsorption and familial renal glycosuria, and briefly report on targeting of SGLTs
for new therapies for diabetes.
I. INTRODUCTION
Sodium-glucose transporters, also known as Na⫹/
glucose cotransporters or symporters (SGLTs), have a
historical place in the field of membrane transport. Half a
century ago it was established that glucose transport
across the small intestine occurred by active transport,
i.e., the sugar could be absorbed uphill against its concentration gradient both in vivo and in vitro, and this
uptake was blocked by metabolic poisons. Nonmetabolized glucose analogs were also actively transported, and
the process was located at the brush-border membrane of
the enterocytes lining the intestine. An explanation for
active transport of glucose, and other molecules, was
completely lacking until Bob Crane proposed the Na⫹/
glucose cotransport hypothesis in 1960 at the Symposium
on Membrane Transport and Metabolism in Prague (94).
In this model, the energy for uphill glucose transport was
provided by the sodium gradient across the brush-border
membrane, and the sodium gradient was maintained by
the Na⫹/K⫹ pump.
The model proposed by Crane is reproduced in Figure 1.
It shows the brush-border membrane of the intestinal
epithelium with the digestive surface and the diffusion
barrier (plasma membrane). Glucose, liberated from dietary sucrose at the digestive surface, is transported
across the plasma membrane by a sodium-glucose carrier
complex. Glucose transport is driven by the inward Na⫹
gradient maintained by the Na⫹ pump. Strophanthidin
inhibits the Na⫹ pump causing the Na⫹ gradient to dissiPhysiol Rev • VOL
pate and remove the driving force for uphill glucose transport. Phlorizin, a plant glucoside, directly inhibits cotransport. This simple scheme accounts for “active” transport
of glucose across the intestinal brush-border membrane
and the requirement for energy input from the cell. The
model remains valid to this day, apart from some minor
details such as the site of phlorizin inhibition (extracellular) and the location of the Na⫹/K⫹ pump (basolateral
membrane). A personal account of Crane’s road to the
Na⫹/glucose cotransport hypothesis is available (26).
The cotransport hypothesis was initially not well received by physiologists, but Peter Mitchell, who was also
a participant at the Prague Symposium, later generalized
the concept to include both cotransport and exchange
(148). He clearly recognized that coupled transport could
be extended to active transport in bacteria, i.e., H⫹/sugar
cotransport. Mitchell defined coupled transport as secondary active transport and coined the term symport that
has been retained by biochemists. Readers interested in a
wider historical perspective are referred to the monograph by Robinson (184).
In the 1960s and 1970s, the Na⫹ cotransport hypothesis was vigorously tested and extended to include the
“active” transport of other molecules and ions, not only in
the intestine and kidney but in such organs as the brain
and thyroid gland (196). The active transport of ions and
molecules in plants and bacteria was also demonstrated
to be due to “symport” with protons as the driving cations.
As work progressed, Na⫹ symporters in bacteria and H⫹
symporters in mammals have been identified. The most
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
735
FIG. 1. The cotransport model drawn by Crane on
August 24, 1960 in Prague. The redrawn figure was included
in the Appendix to the presubmitted paper (23, 25). (Figure
kindly provided by Bob Crane.)
convincing evidence for cotransport came with technical
innovations: 1) introduction of plasma membrane vesicles
(“Kabackasomes,” Ref. 85). Intestinal brush-border membrane vesicles were then used to show Na⫹ gradientdriven accumulation of glucose in the intravesicular
space (76). 2) The use of isolated intestinal cells directly
demonstrated Na⫹-coupled glucose transport (93). These
advances have been fully reviewed in this journal and
elsewhere (24, 27, 75, 92, 195, 196).
Our review begins in 1987 with the cloning of the
intestinal Na⫹/glucose cotransporter (SGLT1) (61), and
here we focus on the biology of human SGLTs in health
and disease. As more transporters were cloned, it was
found that SGLT1 was the first member of a large group of
proteins in the SGLT (SSS) gene family (http://pfam.
sanger.ac.uk). There are 12 members of the human family
(SLC5), and they include Na⫹ cotransporters for sugars,
myo-inositol, iodide, short-chain fatty acids, and choline
(251). A major shock came with solving the crystal structure of vSGLT (48) when we discovered that members of
unrelated gene transporter families share the same basic
core structure, suggesting a common mechanism (1). It
has also gradually emerged that SGLTs are expressed
throughout the body, indicating their importance in organs other than the intestine and kidney, e.g., SGLT1 is
active in specific regions of the brain such as the hippocampus. This review will cover these advances, but it
should be emphasized that this is a rather egocentric view
of the SGLTs.
Physiol Rev • VOL
II. 1987 BC
In the quarter of a century following its publication
and before cloning (BC), the Na⫹/glucose transport hypothesis was extensively tested in the intestine and kidney using in vitro techniques and radioactive tracers (3Hand 14C-labeled sugars as well as 22Na and 24Na) and/or
electrical assays. It was established that the natural sugars D-glucose and D-galactose, and their nonmetabolized
homologs, 3-O-methyl-D-glucoside and ␣-methyl-D-glucopyranoside, were all transported across the brush-border
membrane in a Na⫹-dependent, phlorizin-sensitive manner, but 2-deoxy-D-glucose and fructose were not. Transport was associated with a depolarization of the membrane potential, and the rate of transport was voltage
dependent. Na⫹ was required in the extracellular solutions to drive uphill transport, but Na⫹ could be mimicked
by H⫹ and Li⫹ and not K⫹, Rb⫹, Cs⫹, or choline⫹. In
terms of sugar selectivity, it was established that hexoses
with equatorial -OH groups at carbons 2 and 3 were
transported, but 2-deoxy-D-glucose, 3-deoxy-D-glucose,
D-mannose, and D-xylose were not. Several glucosides and
␤-glucopyranosides were also found to be transported,
e.g., 3-O-methyl-D-glucoside, and ␤-phenyl-D-glucopyranoside but others, notably phlorizin (215), were potent competitive inhibitors. At least in the case of the chick intestine, it was demonstrated that there was a tight stoichiometry between 22Na and 14C sugar transport with a
coupling coefficient of 2.
91 • APRIL 2011 •
www.prv.org
736
WRIGHT, LOO, AND HIRAYAMA
A great deal of effort was directed at unraveling the
kinetics of cotransport, but this resulted in conflicting
models largely due to limitations of the methods. Experimental limitations included the following: 1) flux measurements are prone to error at low specific activities of
the tracers; 2) difficulty of measuring initial rates of transport under well-defined cis- and trans-conditions; 3) uncontrolled membrane potential; 4) presence of other
transport systems (diffusion, uniport); and 5) possibility
of multiple transporters in natural membranes. Even under the best of circumstances, e.g., voltage-clamped membrane vesicles or isolated cells, it was difficult to counter
all these limitations.
Probably the most succinct discussion of the kinetics
of intestinal Na⫹/glucose cotransport was provided by
Stan Schultz in 1985 (195). His model, Figure 2, was a
rapid equilibrium, ordered, six-state scheme where one
external Na⫹ binds first to produce a conformational
change permitting sugar binding and then Na⫹/glucose
cotransport. The empty carrier was modeled as a negatively charged protein (z ⫽ ⫺1) that was sensitive to
membrane potential. Although this simple equilibrium
model requires eight independent parameters that are
difficult to estimate, the general predictions were consistent with the experimental data, e.g., maximum transport
is independent of external Na⫹ concentration, and the
apparent Km for sugar is Na⫹ dependent and voltage
sensitive. While it was clear that the transporter is reversible, little was known about the kinetics of efflux due to
the inherent experimental hurdles at the time. Further
progress was hindered by the difficulty in teasing out the
kinetics of the partial reactions.
The SGLT1 protein was not definitively identified
until 20 years after the cotransport hypothesis was proposed. This was finally accomplished by the Semenza and
Wright labs through the use of azido-phlorizin-photoaffinity labeling, antibodies, and group specific reagents (79,
169, 192). We demonstrated that the lysine reagents phenyl-isothiocyanate (PITC) and fluorescein isothiocyanate
(FITC) labeled the protein in the absence of Na⫹ and
D-glucose. Specific labeling of the transporter was then
achieved by pretreating the membranes with PITC in the
presence of Na⫹ and glucose, washing, and then labeling
with FITC in the presence or absence of Na⫹ and glucose.
The fluorescently labeled protein was then identified as a
73-kDa band on SDS-PAGE. Na⫹ specifically quenched
the fluorescence of FITC bound to a lysine at or near the
glucose binding site, and this was interpreted as a conformation change that permitted sugar binding (169).
During this period evidence emerged that there are at
least two different Na⫹/glucose transporters. First, microperfusion studies in the kidney revealed that the early
proximal tubule absorbed glucose with a Km (2 mM)
higher than in the late proximal tubule (0.5 mM) (5).
Second, it was found that brush-border membrane vesicles prepared from the renal outer cortex and outer medulla have low- (Km 6 mM) and high-affinity (Km 0.3 mM)
transporters (223, 224). The low-affinity SGLT transporter
had an apparent coupling stoichiometry of 1 Na:1 sugar,
whereas for the high-affinity transporter, it was 2 Na:1
sugar. The low-affinity transporter became to be known
as SGLT2, and the high-affinity transporter was determined to be SGLT1 (245). Third, different inherited defects of glucose transport were found in the intestine and
⫹
FIG. 2. A mechanical model for Na -coupled sugar transport. This is a 6-state rapid equilibrium, alternating access model, based on that
proposed by Schultz (195). States 1–3 face outward, and states 4 – 6 face inward. The unloaded
negatively charged carrier (state 1) has low affinity for external sugar before sodium binds.
After external Na⫹ and sugar binding, the tertiary Na⫹/sugar/carrier undergoes a conformational change to present the Na⫹ and sugar binding sites to the cytoplasm where the ligands are
released, either Na⫹ or glucose first. The unloaded carrier (state 6) then undergoes a further
conformation change to reexpose the binding
sites to the external surface of the membrane.
The voltage dependence of Na⫹/sugar cotransport could arise from the electrodiffusion of
Na⫹ in and out of the binding site at each side of
the membrane, and/or the translocation (reorientation) of the charged form of the carrier from
one surface to the other.
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
kidney, i.e., intestinal glucose galactose malabsorption
(GGM) was not associated with a major defect in renal
glucose reabsorption (193), while congenital defects in
renal reabsorption (familial renal glucosuria, FRG) were
not accompanied by defects in intestinal absorption (244,
250).
III. EXPRESSION CLONING
The mid-1980s brought astonishing progress in the
cloning of membrane proteins as exemplified by the successes with the facilitated glucose transporter GLUT1
(150) and the red cell Cl/HCO3 exchanger (97). These
clones were isolated from cDNA libraries by screening
with antibodies and/or synthetic oligonucleotide probes
based on partial amino acid sequences. We were determined to clone SGLT1 and initiated a program to purify
and sequence FITC-labeled SGLT from rabbit brush borders (Brian Peerce), recruit a postdoctoral fellow with
experience in DNA sequencing (Matthias Hediger), and an
undergraduate student (Tyson Ikeda) to develop a heterologous expression system to study the functional properties of clones. Michael Coady, a new graduate student,
also joined the team. Cameron Gunderson, who had just
returned to UCLA from a postdoctoral position at University College in London, advised us to use Xenopus laevis
oocytes as an expression system. Soon thereafter we
established that native oocytes did not exhibit an endogenous sodium glucose cotransporter as determined by
uptakes of ␣-methyl-D-glucopyranoside (␣MDG) in the
presence and absence of Na⫹. However, injection of intestinal poly(A)⫹ RNA into the oocytes increased the rate
of Na⫹-dependent ␣MDG uptake by an order of magnitude compared with H2O-injected control oocytes. This
established that oocytes were a good system to study the
function of cloned SGLTs.
Unfortunately, our attempt to purify and sequence
SGLT1 peptides failed, so by necessity we used transport
assays in oocytes to screen intestinal cDNA libraries. To
simplify the task, Matthias Hediger used preparative gel
electrophoresis to fractionate mRNA, and he isolated an
enriched fraction (2.0 –2.6 kb) containing the mRNA coding for SGLT1 (62). cDNA was synthesized from this
fraction, and a plasmid library was constructed in the
Bluescript expression vector. Synthetic RNA was prepared from cDNA from pools of clones, and this was used
to screen for transport activity in the oocyte expression
assay. One pool of clones gave a positive signal, and this
group was further subdivided until a single clone was
isolated: pMJC424, selected by Mike Coady on 4/24/1987,
increased Na⫹-dependent ␣MDG uptake by more than
1,000-fold (61). Soon thereafter, human SGLT1 was
cloned (64).
Physiol Rev • VOL
737
A. The Intestinal Brush-Border Transporter
Is SGLT1 the intestinal brush-border transporter?
This question was resolved by comparing and contrasting
the properties of the cloned transporter expressed in
oocytes with the Na⫹/glucose cotransporter in the intestinal brush-border membranes (82). The ion specificity,
sugar specificity, and sodium activation plots were similar: neither choline, nor K⫹, nor Li⫹ could efficiently
replace Na⫹ in driving sugar transport; sugar transport
had a Na⫹ Hill coefficient of close to 2; and other hexoses
inhibited ␣MDG transport while mannitol and L-glucose
did not. Furthermore, the properties of the cloned rabbit
and human transporters were similar when expressed in
mammalian cell lines (COS7), insect cells (Sf9 cells), or
bacteria (10, 176, 202, 228). The apparent ␣MDG affinity
(K0.5) was 0.1– 0.5 mM in oocytes, CO7 cells, and Escherichia coli. The kinetics for ␣MDG transport across intestinal brush-border membranes is more complex: apparently with high- and low-affinity systems in brush-border
membrane vesicles (K0.5 values of 0.04 and 0.6 mM, Ref.
82); and an unexplained discrepancy between these apparent affinities and those measured in vivo, 10 –30 mM
for glucose, galactose, and ␣MDG in rats (31). This discrepancy also holds for the apparent inhibitor constant
for phlorizin, 0.2 ␮M for rat SGLT1 in oocytes to 100 ␮M
in vivo (6, 163).
The cloned SGLT1 expressed in oocytes is electrogenic (11, 226). Figure 3A shows that 10 mM ␣MDG
depolarized the membrane potential by 70 mV (from ⫺35
to ⫹35 mV), and under voltage clamp (⫺50 mV), the sugar
FIG. 3. Electrical properties of hSGLT1 expressed in oocytes. The
oocyte was bathed in 100 mM NaCl buffer (in mM: 100 NaCl, 2 KCl, 1
MgCl2, 10 HEPES/Tris, pH 7.5) at 20°C. A: depolarization of the membrane potential after addition of 10 mM ␣MDG. B: inward current
produced by 10 mM ␣MDG when the same oocyte was voltage-clamped
at ⫺50 mV. (From D. D. F. Loo and E. M. Wright, unpublished data.)
91 • APRIL 2011 •
www.prv.org
738
WRIGHT, LOO, AND HIRAYAMA
induced an inward current of 400 nA (Fig. 3B). Neither
membrane depolarization nor inward currents are observed in the absence of Na⫹ in SGLT1 expressing
oocytes or in control oocytes. The electrophysiological
properties of SGLT1 can be then used to explore sugar
and ion specificity.
Figure 4A shows an experiment with hSGLT1 (37).
The addition of 1 mM ␣MDG rapidly produced an inward
current of 800 nA, and this was reversed by washing out
the sugar in a Na⫹-free buffer (choline Cl). Restoring the
external Na⫹ buffer brought the current back to the starting level. No inward sugar-induced currents were observed in the absence of Na⫹ (choline Cl), i.e., the sugarinduced currents are inward Na⫹ currents. The addition
of the ␤-glucoside indican (2 mM) also produced a reversible inward current, but the ␤-glucoside esculin (10 mM)
did not. Phlorizin alone (100 ␮M) did not induce an inward current but rather inhibits a small inward current,
and this suggests that SGLT1 transports a modest
amount of Na⫹ in the absence of sugar (226). There was
no effect of phlorizin in control oocytes or in hSGLT1
oocytes in the absence of Na⫹. The final part of this
experiment shows the second trial with ␣MDG produced an inward current similar to the first and that
esculin partially inhibits the ␣MDG current. This simple
experiment suggests that indican is a substrate for
hSGLT1, while esculin is not but instead behaves as an
inhibitor (37).
An advantage of the electrical assays for SGLT1 sugar
transport kinetics is that radioactive substrates are not
required. This is illustrated for two sugars in Figure 4B,
where we measured the kinetics of both ␣MDG and
3-fluoro-3-deoxy-D-glucose at an external NaCl concentration of 100 mM and a membrane potential of ⫺50 mV (38).
Both sugars have close to the same maximum velocity
(maximum sugar currents of 860 and 775 nA) but different
apparent affinities (K0.5 values of 0.5 and 9 mM). The
␣MDG K0.5 is similar to that obtained with radioactive
tracer uptakes. A further advantage of the electrophysiological assays is that it is possible to obtain a complete
data set on a single oocyte for the kinetics of sugar
transport as a function of the membrane potential and the
external sodium, sugar, and inhibitor concentrations (presented below). Clearly, the magnitude of the maximum
currents depends on the level of SGLT expression that
varies from cell to cell.
Taken together with studies of glucose-galactose malabsorption (see sect. XIVD), these data show that the
clone isolated from the small intestine is in fact the brushborder SGLT1.
B. Stoichiometry of Naⴙ to Glucose Transport
The ability of SGLTs to accumulate sugar is critically
dependent on the stoichiometry of the Na⫹ and sugar
⫹
FIG. 4. A: Na current was measured in a single hSGLT1 injected oocyte in the presence of different glycosides (37). The membrane potential
was clamped at ⫺50 mV. The horizontal line indicated the baseline current in Na⫹ medium in the absence of substrate. The addition of 1 mM ␣MDG
and 2 mM indican induced inward currents, 100 mM Pz reduced the baseline current, and 10 mM Esculin did not produce a current but inhibited
the ␣MDG current by 30%. After the addition of each sugar, the oocyte was washed out in Na⫹-free medium (black box) followed by Na⫹ medium
(blank box). B: kinetics of ␣MDG and 3F3DOglc in a single oocyte expressing hSGLT1 by measuring sugar-induced currents as a function of sugar
concentration (38). Kinetic constants (means ⫾ SE), K0.5, and Imax were obtained by fitting the sugar-dependent currents for each sugar
concentration to the equation, I ⫽ Imax ⫻ [S]/ (K0.5 ⫹ [S]). The apparent affinities for each sugar in each SGLT are summarized in Table 2.
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
fluxes, i.e., [S]i/[S]o ⫽ ([Na⫹]o/[Na⫹]i)n exp(VF/RT)n,
where [S] and [Na⫹] are the intracellular (i) and extracellular (o) sugar and Na⫹ concentrations, respectively; V is
the membrane potential, F is the Faraday constant, R
is the gas constant, T is the absolute temperature, and n
is the coupling coefficient. This relationship gives the
theoretical equilibrium intracellular sugar concentration
for a given Na⫹ electrochemical potential gradient, and it
assumes that there is no sugar metabolism and no other
route of sugar transport into or out of the cell. Thus, for
the same Na⫹ electrochemical potential gradient, e.g.,
V ⫽ ⫺36 mV and [Na⫹]o/[Na⫹]i ⫽ 5, increasing the stoichiometry from 1 to 2 increases the theoretical sugar
concentration ratio by 20-fold (see Ref. 92). An experimental problem in establishing the stoichiometry using
thermodynamics has been the uncertainly in determining
the membrane potential, the internal Na⫹ and sugar concentrations, and sugar metabolism. This has been overcome by measuring reversal potentials with defined ion
and nonmetabolized sugar concentrations on each side of
the membrane (17, 81).
We determined stoichiometry by simultaneously
measuring currents and radioactive tracer fluxes into single oocytes expressing rabbit SGLT1 (139). The experimental protocol is illustrated in Figure 5A, where the
baseline current of an oocyte is recorded in the absence
of sugar and radioactive tracers. The current was then
recorded continuously while superfusing the oocyte with
200 ␮M ␣MDG and 10 mM 22NaCl for 10 min. The integrated sugar-induced currents (charge expressed in
moles) are plotted against the 22Na uptakes in Figure 5B
and give a slope of 1.0 ⫾ 0.1, i.e., the sugar-induced
739
current through SGLT1 is exclusively a Na⫹ current. Similar experiments were carried out to measure 50 –500 ␮M
[14C]␣MDG uptakes, and Figure 5C shows that plotting
charge versus sugar uptakes has a slope of 1.6 ⫾ 0.3. With
a higher Na⫹ concentration (100 mM NaCl at ⫺110 mV),
the Na⫹ to ␣MDG coupling ratio is 1.9 ⫾ 0.1. These results
are in close agreement with reversal potential measurements on hSGLT1 (17, 81) and the radioactive tracer
experiments on chick intestinal cells (93). The stoichiometry for pig SGLT3 is also 2/1 (34).
C. Reversible Transport
Two major approaches have been used to study reverse transport: the first was to express rabbit SGLT1 in
oocytes and employ isolated patch-clamp techniques to
measure the kinetics of Na⫹/glucose outward currents
(47, 191), and the second was to express hSGLT in bacteria, prepare right-side-out and inside-out membrane vesicles, and measure the transport kinetics of sugar uptake
using radioactive tracer techniques (176).
The patch-clamp approach is illustrated in Figure 6A.
Currents were recorded from an isolated patch of oocyte
plasma membrane expressing SGLT1. The pipette solution contained a buffer with 10 mM NaCl and no sugar
(the extracellular solution) while the bath (intracellular
solution) contained 0 –500 mM NaCl and 0 –500 mM sugar.
The sugar-dependent currents across the membrane were
recorded at voltages between ⫺150 and ⫹50 mV (intracellular side with respect to the extracellular side). In this
particular experiment, the current was recorded with an
⫹
FIG. 5. Stoichiometry of Na /glucose cotransport. A: sugar-dependent current (with membrane potential clamped at ⫺70 mV) was continuously
monitored in an oocyte expressing rabbit SGLT1 (139). Baseline current was stable in 10 mM Na⫹ before superfusing 10 mM ␣MDG together with
10 mM 22Na for 10 min (shown by the solid bar) before washing out with 10 mM Na⫹ (without tracer). The sugar-dependent inward charge (Q␣MG),
i.e., integral of the sugar-induced current over 10 min, was ⫺1.4 ⫻ 10⫺4 Coulombs, equivalent to 1,425 pmol of monovalent charge. 22Na
accumulation in this oocyte was 1,398 pmol (having subtracted the mean basal 22Na accumulation over 10 min in control oocytes), yielding a
charge-to-22Na ratio of 1:1. B: charge/22Na stoichiometry for rSGLT1. Sugar-dependent charge (Q␣MG) and 22Na accumulation over 10 min at 10 mM
Na⫹ were simultaneously determined at ⫺70 mV in 19 oocytes expressing rSGLT1. The charge/22Na stoichiometry was 1.0 ⫾ 0.1 (SE). C: Na⫹/␣MDG
stoichiometry for rSGLT1. Q␣MG was compared with the accumulation of 50 –500 ␮M [14C]␣MDG over 1–10 min, at ⫺70 mV. The mean Na⫹/␣MDG
coupling coefficient was 1.6 ⫾ 0.3 at ⫺70 mV.
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
740
WRIGHT, LOO, AND HIRAYAMA
⫹
FIG. 6. Outward Na /sugar cotransport by rabbit SGLT1 in an excised inside-out patch of oocyte membrane (47). A: membrane potential was
clamped at 0 mV, at the time shown by the bar, and 100 mM ␣MDG was added to the external superfusing solution (internal surface of SGLT1). Sugar
induced an upward deflection of the current trace, signifying an outward current (20 nA). Pipette (or external) solution contained the following (in
mM): 10 NaCl, 90 choline Cl, 2 KCl, 1 CaCl2, 1 MgC12, and 10 HEPES, pH 7. 5. Bath (or internal) solution contained the following (in mM): 500 NaCl,
2 KCl, 1 CaCl2, 1 MgCl2, and 10 HEPES, pH 7.2. B: dependence of the outward current on internal [␣MDG] (47). Current records in the same patch
(expressing rabbit SGLT1) as various concentrations of ␣MDG were added to the bath solution. Pipette and bath solutions contained 10 and 500
mM Na⫹ at 0 mV (Vm). C: relationship between the ␣MDG-induced outward current and [␣MDG]i, from the experiment of Fig. 6B. The data followed
a hyperbolic relation with a K0.5 for ␣MDG of 37 ⫾ 5 (SE) mM. Population average was 32 ⫾ 8 mM (n ⫽ 5).
intracellular NaCl concentration of 500 mM at a holding
potential of 0 mV. The addition of 100 mM ␣MDG to the
intracellular solution produced an outward current (intracellular to extracellular current) of 20 pA that reversed on
washing out the sugar. Figure 6B shows the outward
currents produced by 10 –500 mM ␣MDG, and Figure 6C is
the plot of the sugar-induced current against the internal
sugar concentration which yields a K0.5 of 37 mM. As
judged by the currents produced by different sugars at 100
mM, the relative sugar affinities for reverse transport is
␣MDG ⬎ D-galactose ⬎ 3-O-methyl-D-glucoside ⬎ D-glucose. This is quite different from the sugar selectivity for
inward transport, ␣MDG ⬃ D-galactose ⬃ D-glucose ⬎⬎
3-O-methyl-D-glucoside. Phlorizin is also a poor inhibitor
from the cytosolic side of the membrane with a Ki estiPhysiol Rev • VOL
mated to be ⬎1 mM compared with ⬍1 ␮M at the extracellular surface.
The kinetics of forward and reverse sugar transport
by human SGLT1 was also studied by monitoring radioactive ␣MDG uptake into right-side-out and inside-out
membrane vesicles from bacteria expressing hSGLT1
(176). In both cases, incubation buffer contained 100 mM
NaCl and the intravesicular buffer contained 100 mM
K-phosphate. The ␣MDG K0.5 for transport in the forward
direction (extracellular to intracellular) was 0.15 mM
(identical to that for SGLT expressed in oocytes, COS7,
and Sf9 cells), while that for ␣MDG transport in the
reverse direction (intracellular to extracellular) was 56
mM. ␣MDG transport in the forward direction was Na⫹
dependent, phlorizin sensitive (Ki 2 ␮M), and inhibited by
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
10 mM D-glucose, ␣MDG, and D-galactose, but not 2-deoxy-D-glucose, i.e., has the same sugar and phlorizin selectivity as uptakes into eukaryotes expressing hSGLT1.
On the other hand, transport in the reverse direction was
reduced by 100 mM sugar in the order ␣MDG ⬎ D-galactose, 3-O-methyl-D-glucoside ⬎ D-glucose, and there was a
poor affinity for phlorizin, i.e., similar to that observed
with reverse transport in the patch-clamp experiments
(see above). These patch-clamp and vesicle studies show
that transport through SGLT1 is indeed reversible, but the
kinetics and sugar selectivity are asymmetrical. The implication is that there are slight differences in the substrate binding site depending on the direction of transport.
In summary, this preliminary characterization of
the hSGLT1 clones, and those from rabbit (228), rat
(164), mouse (33), and sheep SGLT1 (M. Bing, D. D. F.
Loo, B. A. Hirayama, S. P. Shirazi-Beechey, and E. M.
Wright, unpublished data), is quite consistent with all
that was known about intestinal brush-border Na⫹/
glucose transporter prior to cloning in 1987.
IV. GENES
A. Chromosomal Location
The chromosomal location of hSGLT1 was first assigned to the q11.2 qter region of chromosome 22 using
Southern blots of genomic DNA from a panel of hamsterhuman somatic cell hybrids (60). This was further refined
to the proximal half of band q13.1 by fluorescence in situ
hybridization (FISH) with metaphase chromosomes from
normal subjects and patients with translocations of chromosome 22, e.g., chronic myelogenous leukemia (CML)
(219). hSGLT2 was assigned to chromosome 16 close to
the centromere (240). The human genome project ultimately led to the gene mapping of all six SGLTs (Fig. 7)
beginning with SGLT1 on chromosome 22 (39): SGLT2
16p12-p11; SGLT3 21q22.12; SGLT4 1p32; SGLT5 17p11.2,
and SGLT6 16p12.1 (251).
B. Gene Mapping
The entire SGLT1 gene was initially mapped by cloning the gene from cosmid and ␭ phage clones, restriction
mapping, and sequencing the exon/intron boundaries
(220). This single-copy gene is large with 15 exons spanning 72 kb (Fig. 7). Transcription is under the control of a
TATA box 27 bp upstream of the start codon, and the
minimal promoter, as judged by luciferase reporter assays
in Caco-2 cells (143), is encoded in nucleotides ⫺235/⫹22.
This minimal promoter contains three cis-elements, a
HNF-1, and two GC boxes, which are critical for basal
Physiol Rev • VOL
741
expression and a novel 16-bp element that bind members
of the Sp1 family of proteins that enhance basal expression. At least in sheep, HNF-1 (the HNF-1␣ isoform) appears to be involved in the increase of intestinal SGLT1
gene expression in response to dietary glucose (229).
Interestingly HNF-1␣⫺/⫺ mice do not show any defect in
intestinal glucose absorption, but instead show a renal
defect in glucose reabsorption caused by a reduction in
SGLT2 expression (172).
The organization of all 6 SGLT genes is quite similar,
in having 15 exons, although they span from 8 to 72 kb
(Fig. 7), but in other members of the SLC5 gene family the
coding sequences are found in 14 (NIS), 8 (CHT), or only
1 exon(s) (SMIT1) (251). In SGLT 4 – 6 there are some
indications of alternative splicing, and this may account
for the difficulty in expressing these clones in heterologous expression systems.
C. Human Gene Family (SLC5)
The SGLT family grew larger with the identification
of the renal Na⫹/glucose (SGLT2) (240), the renal Na⫹/
myo-inositol (SMIT1) (104), the thyroid Na⫹/iodide (NIS)
(29), and Na⫹/multivitamin (SMVT) (174). In humans, this
is the SLC5 family (Fig. 8) (253).
All members of the SLC5 family code for 60- to
80-kDa proteins containing 580 –718 amino acids. It is
noteworthy that human SGLT3 is not a transporter but a
glucosensor (35), and this underscores the importance of
functional studies.
When human SGLT1 was first cloned, we found that
it was homologous to the E. coli Na⫹/proline transporter
PutP with a 28% amino acid identity (64). This family of
genes, now known as the sodium solute symporter family
(SSS or SSF), contains hundreds of proteins of pro- and
eukaryotic origins with a common architecture. The SSF
human genes belong to one of 14 members of the Pfam
family clan APC (Cl0062) (PF00474 http://pfam.sanger.
ac.uk). The SSF structural domain proteins share the consensus sequence [GS]-2(2)-[LIY]-x(3)-[LIVMFYWSTAG](7)x(3)-[LIY]-[STAV]-x(2)-G-G-[LMF]-x-[SAP]. There is also a
common motif for the six SGLTs and SMIT1s [RxTxxxxFLAGxxxxWWxxGAS] located on the intracellular loop between transmembrane helices (TM) 1 and 2 (Fig. 1 in Ref.
249). Other families in the APC clan include amino acid and
bicarbonate cotransporters and exchangers. Readers should
be aware that there is another classification system for
membrane transporters where the SGLTs are included in the
electrochemical potential-driven transporters as subclass
2.A.21 (http://www.tcdb.org/tcdb). These two websites,
along with those at the United States National Library of
Medicine (http://www.ncbi.nlm.nih.gov) and GeneCards
(http://www.genecards.org ), provide a rich source of information about the SGLTs.
91 • APRIL 2011 •
www.prv.org
742
WRIGHT, LOO, AND HIRAYAMA
FIG. 7. The organization of human
SGLT1– 6 genes. All SGLT coding sequences are found in 15 exons. (From E.
Turk and E. M. Wright, unpublished
data.)
While we are fairly confident that all members of
the human SGLT (SLC5) gene family have been identified, we recognize that other Na⫹-dependent sugar
transporters may still be hiding in the human genome.
One such novel transporter, NaGLT1, has been cloned
from a rat kidney cDNA library (78). A human homolog
has been identified, KIAA1919, and located on chromosome 6, 6q22. NaGLT1 is a 484-residue protein with
⬍22% amino acid identity with SGLT and GLUT transporters. The gene is mainly expressed in the proximal
tubule at a higher level than SGLT1 and SGLT2. When
expressed in Xenopus laevis oocytes, the protein increased Na⫹-dependent ␣MDG uptake fourfold, and
Physiol Rev • VOL
this was blocked by phlorizin. Neither D-galactose nor
was transported by this low-affinity transporter. In HEK293 cells, NaGLT1 also behaved as a
Na⫹-dependent, phlorizin-sensitive, D-fructose transporter (77).
D-mannose
D. Functional Characterization
Table 1 summarizes the functional properties and
tissue distribution of the six human SGLTs. As discussed
above, SGLT1 transports the natural sugars glucose and
galactose with similar kinetics (K0.5 and Vmax), and this
91 • APRIL 2011 •
www.prv.org
743
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
with unusual sugar selectivity, i.e., it transports D-mannose but not galactose or 3-O-methyl-D-glucoside. The
function of SGLT5 has only been reported in abstract
form (113). When expressed in HEK293 cells it exhibits
Na⫹-dependent, phlorizin-sensitive transport of ␣MDG
and galactose. This gene is exclusively expressed in the
renal cortex. SGLT6 (SMIT2) has the lowest amino acid
identity with SGLT1 (50%), and its preferred substrate is
not glucose but D-chiro-inositol (118). D-Glucose inhibits
with a Ki of 6 mM. Rabbit SGLT6 transports glucose with
a K0.5 of 35 mM (21).
V. EXPRESSION
A. mRNA
FIG. 8. An unrooted phylogenic tree of the 12 human members of
the SLC5 gene family. SGLT6 is also known as SMIT2 (Na⫹/inositol
cotransporter 2). SMIT, sodium myo-inositol; CHT, choline; SMVT, sodium multivitamin; SMCT, sodium monocarboxylic acid; NIS, sodium
iodide cotransporters. [Revised from Wright and Turk (253).]
protein is found abundantly in the brush-border membrane of the small intestine. Relatively little is known
about the functional properties of the kidney SGLT2 as it
is very poorly expressed in heterologous expression systems, but it is a glucose transporter that has poor affinity
for galactose (81, 86, 214). SGLT2 has become a major
drug target for regulating blood glucose levels in diabetes
(see sect. XIVJ). SGLT3 is not a transporter in some
species, e.g., in humans SGLT3 is a glucosensor expressed
in the enteric nervous system and muscle (35). The sensor
does not recognize galactose, but it has a very high affinity
for imino sugars (235). There is one report on SGLT4
(214) showing that it is a low-affinity glucose transporter
TABLE
Analysis of gene expression in human tissues has
been carried out using Northern blots, real-time PCR, and
RNAase protection assays on commercial mRNA samples,
from a very limited pool of Caucasian individuals with
unknown medical histories. A comprehensive study of
SLC gene expression, including SGLT1, -2, -3, -5, and -6,
using these samples has been published (151). Others
have determined expression as the genes were cloned,
e.g., SGLT4 with reference to SGLT1 and -2 (214). In the
PCR experiments, SGLT1 was most abundant in the small
intestine, and significant levels were also found in trachea, kidney, heart, and colon; SGLT2 was exclusively in
kidney; SGLT3 in small intestine; SGLT4 in small intestine
and kidney; SGLT5 in kidney; and SGLT6 in spinal cord,
kidney, and brain. We found similar results in our RNase
protection assays. (Figure 9 shows the profile of SGLT2
mRNA expression.) However, we find SGLT2 mRNA in
cerebellum and low levels in tissues such as heart, salivary gland, liver, and thyroid; SGLT1 mRNA in testis;
SGLT3 in testis, spleen uterus, brain, kidney, and lung;
SGLT4 in liver, brain, and lung; SGLT5 exclusively in
1. SGLT substrates and expression in the human body
Gene
Substrate
K0.5, mM
SGLT1 (SLC5A1)
Cotransporter
SGLT2 (SLC5A2)
Cotransporter
SGLT3 ( SLC5A4)
Glucosensor
SGLT4 (SLC5A9)
Cotransporter
SGLT5 (SLC5A10)
Cotransporter
Glucose, galactose
0.5
0.5
6
NI
20
NI
2
0.15
ND
ND
Glucose
Glucose
Glucose, mannose
Glucose
Galactose
Distribution
Intestine, trachea, kidney, heart, brain, testis, prostate
Kidney, brain, liver, thyroid, muscle, heart
Intestine, testis, uterus, lung, brain, thyroid
Intestine, kidney, liver, brain, lung, trachea, uterus, pancreas
Kidney cortex
Substrate specificity, affinity (K0.5 for ␣MDG), and RNA expression of human SGLT (SLC5) genes are shown. Substrate specificity and ␣MDG
transport were measured using heterologous expression systems (34, 81, 116, 211). RNA distribution is based on mRNase protection assays (M. Bing,
M. G. Martin, and E. M. Wright, unpublished data) and Northern blots (SMIT2). ND, not determined; NI, noninteracting. [Revised from Wright et al.
(246).]
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
744
WRIGHT, LOO, AND HIRAYAMA
FIG. 9. Expression of SGLT2 mRNA in human tissue samples. These
studies were carried out by RNase protection assays on commercial
samples of human mRNA. The autoradiograms were digitized, and the
protected fragments of SGLT2 mRNA were digitized, counted, and expressed as a percentage of that in the tissue expressing the highest
abundance. mRNA extracted from the duodenum and renal cortex was
included as controls. (From M. Bing, E. Turk, M. G. Martin, and E. M.
Wright, unpublished data.)
tibodies and the availability of tissue. Soon after cloning of rabbit SGLT1, a series of polyclonal antibodies
were raised to different SGLT1 peptide sequences and
screened for their specificity on brush-border membranes using Western blots (70, 209). Our criteria for
acceptable antibodies were that 1) antibodies made
against two different peptides are able to recognize the
same single band, and 2) the antigenic peptide had to
block antibody binding. Two antibodies 8792 and 8821
met these criteria in blots of rabbit brush-border membranes. Figure 10 shows a Western blot of mouse intestinal brush-border membranes with a single, broad band of
75 kDa. No other immunoreactive bands were observed in
these blots. We later demonstrated that the broad band
was in part due to N-linked glycosylation of the brushborder SGLT1 (71). The same antibodies recognized
cloned rabbit SGLT1 expressed in Xenopus laevis
oocytes, but two bands are observed representing the
core and unglycosylated protein (70).
It is important to recognize that membrane proteins run faster than expected from their molecular
weight on SDS-PAGE gels with broad bands, presumably due to incomplete denaturing of membrane proteins by SDS. It appears that the apparent molecular
weight of SGLTs can vary with the degree of glycosylation in that SGLT1s from different species range in
size from 69 to 79 kDa (70). Finally, under different
experimental conditions, multimeric SGLT1 bands may
kidney cortex; and SGLT6 in brain, kidney, and small
intestine.
What remains to be determined is the cellular distribution of SGLT mRNAs in the organs and tissues of
the human body, e.g., by in situ hybridization methods.
This has been carried out for kidney, brain, and spinal
cord in animal models (86, 173, 257; Allen Institute
Brain Atlas http://www.brain-map.org/; Ref. 28). Significant levels of SGLT1 and SGLT2 mRNA are detected in
specific regions of the pig, rabbit, rat, and mouse brain,
e.g., pyramidal cells of the hippocampus, and in rat
kidney SGLT2 mRNA is present in S1 segment tubules
in the cortex while SGLT1 mRNA is in the outer stripe
of the outer medulla.
Another source of information about SGLT gene expression in normal human tissues is the EST (expressed
sequence tag) data bases, e.g., www.ncbi.nlm.nih.gov/unigene.
For example, SGLT2 expression is similar to that for mRNA
(Fig. 9), but positives are also reported for testis, placenta,
larynx, pancreas, placenta, and stomach.
B. Proteins
Antibodies have been used to map SGLT expression in human cells and tissues, but so far there are no
comprehensive studies, due to the lack of specific anPhysiol Rev • VOL
FIG. 10. Western blot of mouse intestinal brush-border membranes for SGLT1 using two antipeptide antibodies (no. 8792, residues 402– 419; no. 8821, residues 604 – 615; Ref. 70). Both antibodies
recognize the same 75-kDa protein, and the immunoreactivity was
blocked with the peptides used to raise each antibody. Similar Western blots were obtained against hSGLT1 expressed in Xenopus laevis
oocytes (140). [From Wright et al. (247).]
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
appear on the Western blots when frozen samples are
used.
We have validated the use of our SGLT1 antibodies
for immunolocalization studies, first in Xenopus laevis
oocytes expressing wild-type rabbit SGLT1 and a mutant R427A rabbit SGLT1 that is not inserted into the
plasma membrane (137). Confocal fluorescence microscopy showed 1) no SGLT immunoreactivity in control
oocytes, 2) immunoreactivity was restricted to the
plasma membrane in oocytes expressing wild-type
SGLT1 (within the resolution of light microscopy), and
3) immunoreactivity only appeared just below the
plasma membrane in oocytes expressing the R427 mutant. Freeze-fracture electron microscopy confirmed
that SGLT1 was not in the plasma membrane of control
or oocytes expressing the mutant in contrast to the
5,000 copies/␮m2 in wild-type expressing oocytes. Similar results were observed with hSGLT1 (140). Second,
we have demonstrated that the intensity of the 78-kDa
immunoreactive band in Western blots of sheep brushborder membranes is proportional to the transport activity over two orders of magnitude (201).
Such antibodies have been used to immunolocalize
SGLT1 protein in the rat intestine and kidney (209), rat
and pig brain (173), and rat heart and skeletal muscle
(45). The studies from the Koepsell group can be used
to illustrate the problems that arise when the peptide
antibody recognizes additional bands on Western blots.
Their antibody to residues 582– 600 of rat SGLT1 also
recognized a 40-kDa protein (probably a viral receptor
protein) that led to erroneous conclusions, e.g., that
SGLT1 was expressed in endothelial cells in brain (4).
Commercial antibodies are available for SGLT1 and
other SGLTs, but these have not been carefully evaluated, so published results with these have to be reevaluated. One report has appeared using a SGLT2 antibody
to localize SGLT2 in wild-type and SGLT2⫺/⫺ mice
(227). This study clearly identifies SGLT2 in the brushborder membrane of early proximal tubules.
As yet, the potential to map the distribution of SGLTs
throughout the human body by immunolocalization has
not been realized.
745
half-life (109.8 min). 2-FDG is transported into cells via
GLUTs, where it is trapped after conversion to the
phosphorylated sugar 2-FDG-6-phosphate. Accumulation of 2-FDG-6-phosphate in tissues is followed as a
function of time, and a compartmental analysis is used
to extract rate constants for transport and glucose
phosphorylation (80). 2FDG is a poor substrate for
SGLTs (Table 2), so PET studies with this tracer do not
report cellular glucose uptake via SGLTs.
We and others (12, 30) have developed SGLT specific
PET probes to study the distribution of functional SGLTs
in the human body. Our first generation tracer is ␣-methyl4-deoxy-4-[18F]fluoro-D-glucopyransoside (Me4FDG). This
was designed based on functional studies of sugar selectivity: 1) ␣MDG is a substrate for SGLTs and not GLUTs
(92, 252); 2) ␣MDG is not phosphorylated by hexokinase
(7, 258); 3) 4FDG is a high-affinity substrate for SGLT1
(Table 2) (38); and 4) Me4FDG is a high-affinity substrate
for SGLT1 and -2, but not GLUTs (Table 2) (81; B. A.
Hirayama and E. M. Wright, unpublished data).
PET scans of one adult male subject with 2FDG
and Me4FDG are shown in Figure 11. 2FDG was accumulated in regions of the brain, excreted by the kidneys
into the urinary bladder, and accumulated to a lesser
degree in heart, liver, kidneys, and muscle. These observations agree with the expression of GLUT1 in the
blood-brain barrier, GLUT3 in brain, GLUT2 in liver and
kidney, and GLUT4 in muscle. The excretion of 2FDG
into the urine is also expected as 2FDG is not a substrate for SGLTs, and hence, it is not salvaged from the
glomerular filtrate.
C. PET Imaging
Positron emission tomography (PET) has revolutionized studies of glucose uptake and metabolism in
organs and tissues in the human body in health and
disease using 2-[18F]-2-deoxy-D-glucose (2FDG) as a
tracer (see Ref. 170). 2FDG PET is a safe, noninvasive,
imaging method to monitor glucose uptake and trapping in cells and tissues with high spatial and temporal
resolution (2 mm, s). Fluorine-18 is the preferred positron emitter as it decays with a conveniently short
Physiol Rev • VOL
FIG. 11. The biodistributions of FDG (left) and Me-4FDG (right) on
a 64-yr-old male subject (EMW). 10 mCi of each tracer were injected
intravenously into the subject on separate occasions, and whole body
PET scans were collected at 60 min after intravenous injections. (Note
that each scan is normalized to the highest point, i.e., the two scan scales
are not normalized to each other.) The scans show the lack of brain
uptake of Me-4-[18F]FDG, compared with 2-FDG, and there is no elimination of Me-4-[18F]FDG into the urinary bladder. (From E. M. Wright, A.
Halabi, B. A. Hirayama, V. Kepe, and J. R. Barrio, unpublished data.)
91 • APRIL 2011 •
www.prv.org
746
WRIGHT, LOO, AND HIRAYAMA
FIG. 12. The amino acids sequences of
SGLT1-6 were aligned using Maximum Linkage Clustering. Gaps are indicated by
dashed lines, and conserved residues are
highlighted in red (single letter code or
dashed lines). The location of the 14 putative transmembrane helices (TM⫺1 to
TM13) based on the crystal structure of vSGLT is highlighted in yellow (1, 48). Highlighted boxes identify homologous vSGLT
residues forming the sugar (green) and sodium (dark gray) binding sites and the external and internal gates (green). Note the
general conservation of the ligand coordinating and gate residues in the six SGLT
proteins.
In contrast, Me4FDG did not enter the brain, confirming that it is not a substrate for GLUT1 in the
blood-brain barrier, and did not appear in the urinary
bladder, indicating that it was indeed salvaged from the
glomerular filtrate by SGLTs in the proximal tubule.
Me4FDG was accumulated in kidney, skeletal muscle,
heart, liver, prostate, uterus, and testes, suggesting that
SGLT genes are functional in these tissues (Table 1).
Studies are in progress to determine which SGLTs are
functional in these organs and tissues, and how expression is regulated. So far, in vivo and in vitro assays of
Physiol Rev • VOL
SGLT activity in rat show SGLT activity in discrete
regions of the brain, e.g., hippocampus and cerebral
cortices (258).
VI. PROKARYOTE SGLTS
A few thousand SSF (SSS) genes have been identified
in archea, prokaryote, and eukaryote genomes. The interested reader is referred to the phylogenic tree of 41
diverse members of the gene family (253).
91 • APRIL 2011 •
www.prv.org
747
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
FIG.
Ever since SGLT1 was cloned, we have made efforts to overexpress and purify SGLT1 protein for biochemical and structural studies with limited success
(e.g., Refs. 177, 202). Following the success of others in
the overexpression, purification, and reconstitution of
bacterial symporters, we focused our attention on the
structure and function of the bacterial homolog of
SGLT coded by the SgLs gene of Vibrio parahaemolyticis, vSGLT (218). The protein is smaller than hSGLT1, 543 versus 664 residues, but there is 32% identity
and 75% similarity between the amino acid sequences.
This protein has only one cysteine residue, and this
facilitates cysteine-scanning accessibility measurements. Our strategy was first to express the gene in E.
coli and to characterize vSGLT transport activity in
intact cells, membrane vesicles; purified protein reconstituted into proteoliposomes, and solubilized protein
(216, 218, 230, 254).
We found that vSGLT carried out Na⫹-dependent
transport of D-galactose, but not ␣MDG. Phlorizin inhibited transport but with a poor affinity (Ki ⬎ 1 mM). In
proteoliposomes, the K0.5 for D-galactose transport was
158 ␮M, and the Hill coefficient for Na⫹ was 1, suggesting
a 1:1 coupling between Na⫹ and galactose transport. The
Physiol Rev • VOL
12.—Continued
Na⫹ K0.5 depended on galactose concentration; in Xenopus laevis oocytes, it was 17 mM at 0.06 mM galactose
(116). Kinetic studies with proteopliposomes and purified
protein in detergent further revealed that the system was
ordered with Na⫹ binding before sugar and that Na⫹
binding results in a conformation change that underlies
sugar binding (216, 230).
In summary, the functional properties show that
vSGLT has much in common with hSGLT1, but there are
differences in sugar selectivity and Na⫹-to-sugar transport stoichiometry (1:1 rather than the 2:1 for SGLT1).
This encouraged us to purify this protein for structural
studies (see below).
VII. STRUCTURE
A. Amino Acid Sequences
The SGLT genes code for proteins with 596 – 681
residues. Alternative splicing with SGLT4 – 6 may result in
the predicted amino acid content varying by up to 52
residues (253). Relative to human SGLT1, there is between 50 and 70% identity and 67– 84% similarity in the
91 • APRIL 2011 •
www.prv.org
748
WRIGHT, LOO, AND HIRAYAMA
sequences for SGLT2– 6. The greatest divergence in sequence occurs at the extracellular NH2-terminal domain
and the COOH-terminal third of the proteins. This also
holds for members of the larger SSS family (253). Figure 12
shows an alignment of the SGLT1– 6 amino acid sequences.
B. Secondary Structure
The amino acid sequence of hSGLT1 superimposed
on a 14-transmembrane helix model is shown in Figure 13.
Note that the 14 helices have been renumbered from TM
⫺1 to 13 based on the crystal structure of LeuT structural
family (1). This model was based primarily on N-glycoslyation scanning mutagenesis analysis and computer algorithms, e.g., the neural network algorithm to predict
membrane spans (217, 221). Additional experimental approaches used by us and others included antibody recognition of polypeptide epitopes and labeling of cysteine
mutants with “impermeant” alkylating reagents such as
rhodamine maleimide and charged methanethiosulfonates (MTSET, MTSES). While our model is now generally
accepted, there is disagreement about the location of the
90-residue hydrophilic domain between TMs 12 and 13. A
succinct review of experimental evidence for the topol-
ogy of the COOH-terminal domain is contained in a paper
by Gagnon et al. (53). In essence, the controversy stems
from studies on the accessibility of the large, very hydrophilic loop to hydrophilic reagents and antibodies in the
extracellular compartment, raising the possibility of a
reentrant loop between TMs 12 and 13. This was apparently supported by a proteomic study of trypsin digests of
hSGLT1 in proteoliposomes (102), but we think that the
results are ambiguous as hSGLT1 was most likely reconstituted into the liposomes in both orientations.
Results in support of the 14 TM model (Fig. 13)
include experiments on vSGLT. The secondary structure
model for vSGLT closely resembles that for hSGLT1 but
with shorter hydrophilic loops between the transmembrane domains (221). The NH2-terminal of vSGLT was
shown to be extracellular by electrospray ionization mass
spectrometry (ESI-MS) of purified vSGLT: the NH2 terminal retains its formylmethionine that would normally be
excised in the cytoplasm (218). The crystal structure of
vSGLT (48) showed that the protein has 14 transmembrane helices and that the hydrophilic loop between TM12
and 13 remains in the cytoplasm. TM13 lies outside the
core of the structure in both the native protein and that
with an additional COOH-terminal TM helix (glycophorin
A). In Figure 13, we have superimposed the helices in the
FIG. 13. Secondary structure model of hSGLT1 (217). This model shows the sequence of the 664 residues arranged in 14 transmembrane helices
with both the NH2 and COOH termini facing the extracellular side of the plasma membrane. A single N-glycosylation site occurs at Asn (N) 248.
Highlighted are the locations of the helical domains based on the vSGLT structure (48). The numbering of the TMs has been revised to conform with
the LeuT structural fold to allow easy comparisons between structural family members, i.e., TM1 through TM13 (1).
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
vSGLT crystal structure on the hSGLT1 secondary structure model to highlight the agreement between the model
and the crystal structure. Differences between the predicted secondary structure and the crystal structure arise
mainly due to assumptions in modeling of the length and
angle of helical segments crossing the membrane. For
example, the results of cysteine scanning studies on rabbit SGLT1 suggesting a helical loop in the extracellular
loop between TM3 and TM4 have been reinterpreted as
part of TM3 in light of the vSGLT crystal structure (122).
It remains to be determined why the hSGLT loop between
TM12 and 13 is accessible to extracellular hydrophilic
reagents.
Apart from N-linked glycosylation at N248, there is
little direct evidence for other secondary modifications of
SGLT1. N-linked glycosylation is not required for functional expression (63, 71). There is no evidence for Olinked glycosylation (71). Analysis of potential secondary
modification of the SGLTs using PROSITE at PredictProtein (www.predictprotein.org) predicts motifs for cAMP,
PKC, CK2, and Tyr phosphorylation and myristoylation
sites. The importance of these has yet to be tested by
direct experiment (the regulation of SGLT1 expression by
kinases will be discussed below).
C. Monomer
Freeze-fracture studies of hSGLT1 expressed in
oocyte plasma membranes and purified vSGLT reconstituted in proteoliposomes clearly demonstrated that both
are fully functional as monomeric proteins (46, 218). This
conclusion was based on the cross-sectional area of the
proteins within the plasma membrane relative to those of
membrane proteins of known structure that were also
analyzed by freeze-fracture electron microscopy in oocyte
membranes. The dimensions of the vSGLT electron microscopic images, corrected for the thickness of carbon/
platinum coating, also agreed with the dimensions of the
vSGLT crystal structure. One unresolved question is the
discrepancy between these electron microscopic images
of hSGLT1 and the radiation inactivation analysis of transport in rabbit brush-border membranes, suggesting that
SGLT1 is a homotetramer (205).
Hermann Koepsell and colleagues (106) cloned a
67-kDa protein RS1 that was initially claimed to be a
regulatory subunit, but subsequent studies demonstrated that RS1 was not in the plasma membrane but
instead participated in the transcription and posttranslational trafficking of SGLT1 and other transporters
(101, 182, 232, 233).
D. Protein
The major advantages of using vSGLT for structure/
function studies are as follows: 1) vSGLT only contains
Physiol Rev • VOL
749
one cysteine residue, and this is not required for full
functional activity (254). Thus cysteine scanning accessibility measurements can be made on a cysteine-less background (230). 2) vSGLT mutants expressed in bacteria do
not suffer from trafficking defects observed in eukaryotes, and 3) it is straightforward to isolate vSGLT
mutants from bacterial expression systems for biochemical studies. In 1999, Eric Turk took on the challenge to
produce vSGLT protein for structural studies.
Our strategy to produce and purify the transporter
(vSGLT) was to construct an expression plasmid (VNH6A) by
inserting the full SgLs coding region into the pBAD18
vector and appending a COOH-terminal HIS tag for metalchelate chromatography (218). To facilitate optimization
of vSGLT protein expression and purification, we also
constructed a vSGLT-green fluorescence protein (GFP)
fusion plasmid where a 15th TM (glycophorin) was used
to locate GFP in the cytoplasm (VNGFPH6). Both constructs were expressed in XL1Blue cells after induction
with L-arabinose. Membrane vesicles were prepared by
standard methods, and the protein solublization and purification was followed by monitoring GFP fluorescence.
Both the vSGLT and vSGLT-GFP proteins were almost
exclusively expressed in the bacterial plasma membrane,
and each was fully active in cell and membrane vesicle
transport assays (218, 254). The proteins were purified to
homogeneity by metal-chelate and size exclusion chromatography as judged by SDS-PAGE and mass spectrometry.
ESI-MS confirmed the correct mass of the proteins to
within 0.01% (218). Currently, we obtain ⬎5 mg of vSGLT
protein from 10 liters of cultured cells. Biochemical,
transport, and fluorescence assays show that purified
vSGLT protein is fully active when incorporated into liposomes (48, 216, 218, 230, 231). Using similar strategies,
we and others have had success in purifying hSGLT1 from
E. coli and Pichia pastoris (177, 225).
E. Circular Dichroism
Given that vSGLT was fully functional when solubilized in detergent, we examined the secondary structure
of the protein using ultraviolet (UV) circular dichroism
(CD) (216). The fusion protein vSGLT-GFP was included
as a control as the crystal structure of GFP is known (50%
␤-strand and ⬍2% ␣-helix). A large crystallographic reference set of proteins was available, and three algorithms
(CONTINLL, CDSSTR, and SELCON3) were used to predict the helical content of membrane proteins. For example, CD measurements accurately predicted the helical
content of lactose permease (85%, compared with 86% in
the crystal). Our CD spectra predicted 82– 89% ␣-helical
content in vSGLT and 60% in the vSGLT-GFP fusion protein. This decrease with vSGLT-GFP was anticipated
given the low helical content of GFP. We further estimate
91 • APRIL 2011 •
www.prv.org
750
WRIGHT, LOO, AND HIRAYAMA
that 27–33% of the total helical content in vSGLT occurs
outside the plasma membrane domain (216). Again, the
CD predicted secondary structure is quite consistent with
the subsequent vSGLT crystal structure.
D-Galactose changed the near UV CD, 250 –300 nm, of
vSGLT in the presence of Na⫹, but not in K⫹ (216). This,
and the galactose-induced change in tryptophan fluorescence (231), probably reflects conformational
changes at a tryptophan residue at or near the galactose
binding site. A tryptophan was found flanking the Y263
involved in stacking with galactose in the sugar binding
site (48).
Attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy was also used to examine
the structure of vSGLT (112), but the results suggested
that only 35% ␣-helical content in the absence of ligands.
This secondary structure profile was unexpected, but it is
now known that FTIR methods have a tendency to predict
␤-sheet content in ␣-helical membrane proteins where
none exists, e.g., aquaporin 1 and lactose permease. The
FTIR studies do show that there is limited extent of
hydrogen-deuterium (H/D) exchange in vSGLT, 40 –50% in
2 h compared with 80% for lactose permease and GLUT1.
These results are interpreted as showing that there is
limited water access to vSGLT compared with that found
for the H⫹/lactose symporter. Na⫹ and D-galactose each
reduced H/D exchange in vSGLT, and this suggested that
the ligands caused a compaction of the overall protein
structure.
F. Crystal Structure
The first structure of vSGLT was solved in collaboration with Jeff Abramson and his team (48). The
structural model was refined to 2.7 Å, and all residues,
apart from those in TM⫺1 and 36 disordered residues in
two cytoplasmic loop regions, were assigned. Figure 14A contains a topology model of the crystal structure, and
Figure 14B shows a side view of the model in the
membrane plane.
As predicted, the structure contains 14 TMs with
the NH2 and COOH termini on the same side of the
membrane. A single galactose molecule was found in
the center of the protein, where it was occluded from
both the extra- and intracellular compartments. Notable features not anticipated from the amino acid sequence and the analyses of secondary structure were as
follows: 1) the inverted topology repeat (TM1-TM5 and
TM6 –10) that forms the core of the structure (there is
FIG. 14. The structure of vSGLT. A: topology model showing the 14 TM from the NH2 terminal (TM-1) to the COOH terminal (TM13). The blue
and red trapeziums represent the inverted topology of TM1-TM5 and TM6-TM10. B: a side view of the 3-dimensional structure viewed from the
membrane plane. The location of the bound sugar is shown as black and red spheres. Residues involved in sugar binding and gating and Na⫹ binding
are shown on TM as circles. The two discontinuities represent the disordered regions of the protein. [A and B redrawn from Faham et al. (48).] C:
ligand binding sites in vSGLT. An overview of the galactose and Na⫹ binding sites. The sugar is occluded from the exterior by external and internal
hydrophobic gates.
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
no amino acid sequence homology between the repeats,
but the structures can be superimposed with a RMSD of
3.9 Å); 2) the two discontinuous membrane helices,
TM1 and TM6, which lie at the interface between the
two inverted repeats (these two discontinuous helices
are at the core of the sugar binding site); 3) the large
tilts in TM3 and TM8; 4) the variation in the length of
the TM helices; and 5) the presence of helical structures in the extra- and intracellular loops. The total
helical content of the transmembrane domains and the
extracellular loops (72%) was anticipated by the CD
analysis (216). The structure contains a central group
of seven helices (TM1, TM2, TM3, TM6, TM7, TM8, and
TM10) that are supported by a ring of other helices.
Finally, the cytoplamic halves of TM1, TM2, TM5, TM7,
and TM9 form a large hydrophilic cavity that extends
from the inner gate of the sugar binding site (Y263) to
the cytoplasm.
An overview of the location of the galactose and Na⫹
binding sites and the intra- and extracellular gates is
provided in Figure 14C. A close-up view of the galactose
binding site from the extracellular surface in Figure 15A
shows the position of the gates, M73, Y87, F424, and Y263
that block sugar entry and exit. The outer gate residues
are removed from the external view in Figure 15B to
highlight the residues coordinating with galactose. Hydrogen bonds from the residues coordinating with D-galactose include Q428, Q69, E88, K294, S91, and N260, and the
pyranose ring is stacked against the inner gate residue
Y263. The putative Na⫹ binding site, ⬃10 Å away from the
sugar binding site, has coordinating residues S354, S355
(TM8), and the carbonyl oxygens of A62, I65 (TM1), and
A361 (TM8) (see Fig. 16B below). The functional importance of the sugar and sodium coordinating residues has
been confirmed by transport assays on the mutated proteins, Q69A, E88A, K294A, Q428A, and S365A, reconstituted into proteoliposomes (48).
Both the bound galactose and Na⫹ binding sites
face the cytoplasmic aqueous vestibule. Galactose is
prevented from entering the vestibule by the inner gate
tyrosine (Y263), and Na⫹ is held in place by the coordinating residues on TM1 and TM8 (see below).
There are additional helical structures in the hydrophilic loops connecting the transmembrane helices.
On the intracellular surface, a short helix (IL2) between
TM2 and TM3 lies on the outer edge of the hydrophilic
cavity leading to the sugar binding site, and on the
extracellular surface one of two helices (EL7b) forms
extensive contacts with TM1 and TM3. An additional
extracellular helix (EL5) between TM5 and TM6 connects the two inverted repeats, and in hSGLT1, this
loop contains the N-linked glycosylation site (see Fig.
13). The hydrophilic helices are predicted to play roles
in the conformational changes that underlie coupling of
Na⫹ and sugar transport.
Physiol Rev • VOL
751
FIG. 15. A: view of the galactose binding site from the external
surface showing the external gates. B: the galactose binding site. The
external gates were removed for clarity. The tyrosine residue below the
sugar, Y263, is shown in gray spheres. The coordinating residues and
bonds to glucose are indicated by dotted lines. [Redrawn from Faham et
al. (48).]
vSGLT is constructed of two precisely assembled
halves, so it is amazing that these assemble properly to
form a fully functional transporter when they are expressed in the same cell under two different promoters,
the NH2-terminal half (residues 1–279) and the COOHterminal half (residues 280 –543) (254).
G. LeuT Superfamily
The core structure of vSGLT (TM1-TM5 and TM6TM10) is virtually identical to the core structure of LeuT,
an unrelated transporter in the neurotransmitter sodium
cotransporter gene family (NSS) (48, 256). This was soon
followed by reports that two other unrelated sodium
cotransporters, Mhp1 and BetP, share the same core
structure (183, 241). The core structure of all four pro-
91 • APRIL 2011 •
www.prv.org
752
WRIGHT, LOO, AND HIRAYAMA
Two putative Na⫹ binding sites have been identified in
LeuT, Na1, and Na2 (256). By homology, these two sites
were also found in BetP, and one, Na2, was identified in
vSGLT and Mhp1. Na1 overlaps with the substrate binding
sites in LeuT and BetP, where the cation is coordinated with
the carboxyl group of the substrate. In contrast, Na2 is
formed by two conserved serine residues and backbone
carbonyl oxygens (Fig. 16B). The presence of one or two
sodium binding sites is correlated with the stoichiometry of
transport.
The vSGLT2 Na2 site appeared more open than that
in LeuT (Fig. 16B), suggesting a different functional state.
This was supported by molecular dynamic simulations:
Na⫹ was stably bound to the Na2 site in LeuT, whereas in
vSGLT Na⫹ rapidly escaped down the aqueous vestibule
into the cytoplasm (117, 238). More recently it was reported
that the LeuT structural motif extends to a cation-independent arginine/agmatine antiporter (AdiC) in the APC gene
family (50, 54). Na1 and Na2 sites were not found in this
structure, and this validates the functional significance of
the Na1 and Na2 sites in the sodium cotransporters. As yet,
there is no information about the location of the second
Na⫹-binding site in mammalian hSGLT1.
Although it is impossible to deduce transport mechanism from a single structure, it is feasible to gain insight
into the structural rearrangements by examining the
structures of the other transporters (see sect. XII).
FIG. 16. The LeuT structural family. A: a structural alignment of
TM1 and TM6 and the substrate binding sites for vSGLT (red, galactose),
BetP (glycine betaine, blue), LeuT (leucine, green), and Mhp1 (benzylhydantoin, cyan). B: alignment of the Na2 sites on TM1 and TM8.
[Redrawn from Abramson and Wright (1).]
teins can be superimposed (RMSDs 3.8 – 4.5 Å), and as
with vSGLT, the inverted repeats in each protein can be
superimposed even though there is no amino acid sequence similarity. Figure 16A shows that the substrate
binding sites have a common location approximately halfway across the membrane. While the substrate binding
sites in the four proteins share a common location, the
substrate specificities are determined by different coordinating residues.
The four proteins were found to be in slightly different
conformations: LeuT and Mhp1 in an outward occluded
conformation with aqueous vestibule extending from the
substrate binding site to the extracellular surface; vSGLT in
an inward occluded conformation with a hydrophilic vestibule leading from the sugar binding site to the cytoplasm;
and BetP in an intermediate conformation with no aqueous
vestibules leading to or from the betaine binding site (see
sect. XII for further discussion). In each structure, additional TM helices may be present at the NH2 or COOH
termini, and these may or may not be important for function.
Physiol Rev • VOL
VIII. SUGAR SELECTIVITY
Since 1987 the kinetics and specificity of SGLT isoforms have been studied using electrophysiological techniques on cloned rabbit, rat, mouse, or human SGLT1,
SGLT2, and SGLT3 (33, 34, 37, 38, 67, 81, 86, 125, 235).
Electrophysiological methods offer unique advantages:
1) they enable accurate determination of apparent affinities
(K0.5) from submicromolar to ⬎100 mM; 2) measurement
of K0.5 and maximum rates of transport (Imax) for many
different substrates on the same cell (see Fig. 4); 3) allow
discrimination between substrates and competitive inhibitors (Fig. 4); and 4) determine inhibitor kinetics in the
absence of substrates from leak currents (Fig. 4) and
pre-steady-state charge movements (Fig. 20) (66).
A. Monosaccharides
Table 2 summarizes estimates of apparent sugar affinities (K0.5) for the cloned Na⫹/sugar cotransporters
hSGLTs 1– 4, pig SGLT3, and SMIT1–2 from dog and rat.
All transport or bind D-glucose and ␣MDG and are inhibited by phlorizin, but the affinity of the Na⫹/myo-inositol
(SMIT) cotransporters for glucose are 1–2 orders of magnitude lower than hSGLT1. The most essential requirement for substrate interaction and transport by SGLTs is
91 • APRIL 2011 •
www.prv.org
753
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
TABLE
2. Sugar selectivities of SGLT 1– 6
Apparent Affinity, mM
Sugar
hSGLT1
hSGLT2
hSGLT3
pSGLT3
hSGLT4
␣MDG
D-Glucose
D-Galactose
1DOG
2DOG
2F2DOG
Glucosamine
3DOG
3F3DOG
4DOG
4F4DOG
4F4DOGal
5ThioGlc
6DOG
6F6DOG
3-OMG
3-O-benzyl-glc
D-Mannose
D-Fructose
Myo-inositol
D-Chiro-inositol
L-Fucose
L-Glucose
L-Xylose
D-Xylose
D-Allose
D-Fucose
␤MDG
1-Deoxynojirimycin
1-Deoxygalactonijirimycin
N-ethyl-1-deoxynojirimycin
Miglitol
Miglustat
1-Deoxynojirimycin-1sulfonic acid
0.54,5 (4**18)
0.5¶1,4,5 (2**18)
1¶1,4,5 (6**18)
105
⬎1005
⬎⬎10019
⬎⬎10019
⬎1007
95
0.47
0.075
1.37
45
35
⬃519
3.3-64,5
3017
NI1
NI1
⬃5003
4.811,18 (6**18)
5**18
⬎10018
217,15
194,7
NI4,7,5
437
NI7,13
45,13
85,13
⬎2005
265
NI5,13
2.61
7.7¶1
⬎100¶1
⬎30*1
⬎30*1
503
⬎503
⬎⬎503
⬎⬎5020
362
⬎⬎5020
⬎⬎503
NI20
NI*1
⬎⬎503
NI20
NI7
cSMIT 1
rtSMIT 2
NI15
⬎507
177
NI7
17*5
45
⬎507
⬎100¶1
⬎100¶1
⬎⬎1003
⬎⬎1003
⬃503
⬃503
⬎⬎10018
⬃503
0.55
NI7
NI7
NI7
NI7
NI7
NI7
NI13
NI4,5,13
NI13
NI15
NI13
NI13
NI13
NI13
NI13
NI13
75
0.151
⬃1001
0.0503
0.272
0.312
⬃503
⬎503
⬃503
⬎⬎503
NI2,20
⬎⬎5020
⬎⬎5020
⬃5020
⬎⬎503
NI20
0.0047
117
0.0037
0.0037
0.00057
0.0097
*Ki, or estimate of inhibition. **Measured at 37oC. ¶Value estimated from graph in paper. NI, no interaction. §EC50. ‡Kd. #rbSGLT1. References
are as follows: hSGLT4: 1Ref. 214; hSGLT1: 4Ref. 37, 5Ref. 38, 6Ref. 66, 17Ref. 67, 18 Ref. 81, 19Hirayama and Wright, unpublished data; rat SMIT2: 2Ref.
3,15Ref. 21; dog SMIT1: 3Ref. 58; hSGLT3: 7Ref. 235, 15Ref. 35, 13Ref. 138; hSGLT2: 8Ref. 147, 9Ref. 156, 10 Ref. 255, 11Ref. 159, 12Ref. 86; rbSGLT1:14Ref.
136; pSGLT3: 16Ref. 234.
that the sugar must be a pyranose and cyclic polyhydroxy
alcohols are noninteracting. Removal of the one, two,
three, or six equatorial hydroxyls reduces the apparent
affinity for hSGLT1 by a factor of 5 to ⬎200 (1-deoxy-,
2-deoxy-, 3-deoxy-, and 6-deoxyglucose). The O1 in hSGLT3
is of minor importance as removal reduces the apparent
affinity by only a factor of 2. The 4 hydroxyl is not essential for hSGLT1, but replacement with fluorine (4F4DOG)
increases apparent affinity by a factor of 10. Adding F to
2-deoxyglucose (2F2DOG) does not restore the apparent
affinity, whereas in the case of 3-deoxyglucose the affinity
is partially restored (K0.5 decreases from ⬎100 to 9 mM).
The SGLT isoforms 2 and 3 are more selective at the 4
position in that they have poor affinities for D-galactose
relative to D-glucose. Oxygen 5, in the ring, is essential for
recognition by hSGLT1, as replacement by sulfur is detrimental, and substitution by nitrogen abolishes binding
completely. In contrast, imino sugars, e.g., 1-deoxynojirimycin, are the preferred ligand for SGLT3 with micromoPhysiol Rev • VOL
lar affinities. At positions 2 and 3, the hydroxyl must be in
the equatorial position as D-mannose and D-fucose do not
interact with SGLT1, -2, or -3, but mannose is accepted by
SGLT4.
Despite these differences in sugar affinity and selectivity, the sugar coordinating residues are generally
conserved in the SGLT family (Table in Fig. 12). One
significant exception is in SGLT3, where glutamic acid
replaces glutamine at residue 457. Mutation of this
residue, E457Q-SGLT3, resulted in a SGLT1-like phenotype (9), apart from a low galactose affinity (K0.5 ⬎100
mM). In contrast, mutation of 457 in SGLT1, Q457E,
partially produced the SGLT3 phenotype; glucose was
still transported, but the coupling between sugar and
Na⫹ transport was not tight (38).
From studies of hSGLT1 using “mutated sugars”(38,
67, 235), we dissected the essential interactions of sugar
recognition (Table 3). All of the D-glucose equatorial OH
groups and pyranose oxygen are predicted to be involved
91 • APRIL 2011 •
www.prv.org
754
WRIGHT, LOO, AND HIRAYAMA
TABLE 3. Predicted interactions between glucose and
SGLT1
Sugar
Interaction
O1
O2
H-bond donor from protein
H-bond acceptor and donor, with highly constrained
side chain
H-bond donor
Fluorine substitution increased affinity 10-fold, H-bond
not required
Accepts H from protein only; nitrogen substitution
absolutely unacceptable
H-bond acceptor, but not essential; C5 methylene
required
Stacking with an aromatic side chain
O3
O4
O5
O6
Plane
The K0.5 values of a series of sugars in which the hydroxyls and
pyranose oxygen were removed or substituted for other elements were
determined. Comparison to the K0.5 for glucose was then used to propose the pattern and importance of hydrogen bonding involved in recognition. Glucose has hydrophobic surfaces due to the axial orientation
of its hydrogens in the plane of the ring.
in hydrogen bonding to the protein. Based on similarities
to sugar binding proteins, such as the lectins, we also
expected to see a hydrophobic stacking interaction with
an aromatic residue (178, 211). We hypothesized that
Q457 of hSGLT1 interacts with O1, O5, and O6; T460
interacts with O4; H-bonds to and from an unspecified
residue with O2 and a hydrophobic stacking interaction
with an aromatic side chain(38).
How do these predictions compare with those found
in the vSGLT crystal structure (48)? The occluded sugar
binding site in the crystal is dehydrated, and the sugar is
coordinated by H-bonds with polar side chains and stacking of the hydrophobic face of the sugar with tyrosine 263
(Table 4 and Fig. 15B). The coordinating residues are
constrained by adjacent side chains to provide a precise
spatial arrangement for sugar identification.
hSGLT1 has 32% amino acid identity (60% similarity)
with vSGLT, enabling us to model hSGLT1 based on the
vSGLT crystal coordinates. How do the sugar interactions
in the threaded hSGLT1 model compare with the predic-
TABLE
tions based on experiment? Figure 17 shows the model of
glucose binding to hSGLT1. In general, the interactions
between glucose and the protein are accounted for, apart
from a missing O1 interaction. Note that mutation of
Q457C results in a 12-fold reduction in sugar affinity and
that alkylation of this cysteine blocks sugar transport (38,
129). Experiments in progress have confirmed the importance of all the putative sugar coordinating residues (M.
Sala-Rabanal, D. D. F. Loo, B. A. Hirayama, and E. M.
Wright, unpublished data).
B. Glucosides
It has long been known that a variety of glucosides
are transported by SGLTs. Studies from this and many
other groups (e.g., Refs. 2, 37, 107, 136, 149) clearly show
that SGLT1s accept ␤-aromatic and hydrophobic glucopyranosides, with affinity and transportability (Vmax) determined by the groups decorating the aglycone. For example, phenyl-␤-D-glucose is transported by rabbit SGLT1,
but phenyl-␣-D-glucose is noninteracting, i.e., neither a
substrate nor an inhibitor (136). Adding a para-hydroxyl
or amino group to phenyl-␤-D-glucose improves affinity,
whereas a NO2 creates an inhibitor. So the sugar binding
site and translocation pathway is able to accept a large
substituent to the pyranose ring, if it has the right characteristics. In addition to glucopyranosides, glucosides
such as 3-O-methyl- and 3-O-benzyl glucoside are transported by hSGLT1 albeit with a lower affinity than glucose
(Table 5). Large glucosides (up to 20 ⫻ 12 ⫻ 5 Å) are
therefore transported through SGLT1, and this obviously
requires large conformational changes in the protein.
These studies also suggest that there is a series of
“selectivity filters” along the sugar translocation pathway
through the protein and that the characteristics of these
filters can be different from those in the binding site, and
also vary among members of the SGLT family. For example, in rabbit SGLT1, adding a para-hydroxyl to phenyl-␤D-glucose increased both affinity (3 times) and transport-
4. Protein-sugar interactions in vSGLT
Sugar
Interaction
O1
O2
O3
Q69 supplies the essential H-bond
Q69, E88, and K294 donate and accept; E88 and K294 interact with each other to stabilize their location
E88, K294, and S91 donate and accept; position of S91 is constrained by interactions with multiple
residues: W264, N260, and the backbone carbonyl from Y87; E88 and K294 stabilize each other
S91 and N260 donate and accept; position of residues constrained by side-chain interactions
Q428 donates the H-bond
Q428 can donate and accept; the requirement for the 6C may be for distance to the residue from O6 or
hydrophobic interaction with Y263
Stacking with Y263
O4
O5
O6
Plane
The crystal structure of vSGLT contains a bound galactose and reveals the amino acids involved in sugar recognition. These sugar-protein bonds
and interactions between side chains are summarized. We note that the crystal structure is an occluded state and may not represent the initial sugar
binding site.
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
FIG. 17. Glucose binding to hSGLT (homology model). The primary
sequence of hSGLT1 was aligned on that of vSGLT, and this was then
superimposed on the vSGLT crystal structure to create a “threaded”
model structure of hSGLT1 with glucose replacing galactose in the
binding site: O1, no interaction observed; O2 N78, H83 can donate and
accept; O3, E102 can donate and accept; O4, T287, W291 can donate and
accept; O5, Q457 can donate; O6, Q457 donate and accept, possible
hydrophobic with T290; and hydrophobic stacking with Y290. (From V.
Chaptal, B. A. Hirayama, J. Abramson, and E. M. Wright, unpublished
data.)
ability (1.5 times) (136), but in hSGLT1 there was no
change in affinity but transportability increased threefold.
Indican (3-indolyl-␤-D-glucopyranoside) is a high-affinity
(60 ␮M) transported substrate, but it is only transported
␣MDG
at 14% of the maximal rate for ␣MDG (Imax
) in hSGLT1.
It is also a substrate (0.9 mM) in pig SGLT3, but here it is
␣MDG
transported at 80% of Imax
(37). A simple modification
can change transported indolyl glucosides from substrates into an inhibitor. When bromine is added at the
5-position of the substrate 3-indolyl-␤-D-galactopyranoside, 5Br-3-indolyl-␤-D-galactopyranoside becomes an inhibitor (66). Likewise, engineering these transporters may
convert substrates into inhibitors, and vice versa, e.g., a
pig SGLT1/SGLT3 chimera transports phlorizin with a K0.5
of 4.5 ␮M and a maximum transport rate comparable to
that for glucose(161).
C. Inhibitors
The classic competitive inhibitor of SGLTs is phlorizin {Fig. 18A, natural dihydrochalcone glucoside (1-[2-␤D-glucopyranosyloxy)-4,6-dihydroxyphenyl]-3-(4-hydroxyphenyl)-1-propanone} (see an excellent review by Ehrenkranz et al., Ref. 44). Phlorizin is a ␤-glucoside comprised
of two aromatic rings (A and B) joined by an alkyl spacer
of three carbons. As with the aglycone phloretin, it undergoes a keto-enol tautomerization (52, 242), with the
keto-form being the higher affinity configuration (32, 66).
The phlorizin Ki for hSGLT1 is 200 –300 nM, but the
Ki varies among different isoforms, e.g., rat SGLT1 Ki ⫽
Physiol Rev • VOL
755
12 nM and rabbit SGLT1 Ki ⫽ 760 nM (66, 69, 159, 164).
hSGLT2 has an order of magnitude higher affinity for
phlorizin (Ki ⫽ 10 –39 nM, Refs. 81, 88, 159), despite its
10-fold lower apparent affinity for glucose (6 vs. 0.5 mM).
The flexible structure of phlorizin makes predictions
about its three-dimensional structure in the binding site
uncertain (Fig. 18A). First we assume that since phlorizin
is a competitive inhibitor, the glucose is bound at the
same site as the free glucose: this is reasonable given that
the hSGLT1 Ki for phlorizin, 200 nM, is 250 times lower
than that for the aglycone phloretin, 50 ␮M (66). We
propose that the B-ring is canted 30° to the plane of the
sugar as for phenyl-␤-D-glucopyranose. Possible configurations of the A-ring range from that for the X-ray crystal
structure, an extended conformation with the aglycone in
roughly the same plane as the pyranose ring (Cambridge
Structural Database code CEWWAC), to that from solution NMR studies and computer conformational searches,
which predict that that A-ring folds back over the B-ring
(161, 242). The energetic differences between the different conformations are small, so it is not clear how realistically computational methods can be employed to predict the bound conformation.
Nevertheless, analysis of the physical characteristics
of the aglycones combined with their kinetics (Table 5)
allowed us to create a pharmacophore model of the inhibitor binding site, and predict interactions of the aglycone with SGLT1 that determined binding (Fig. 18B) (66).
We predicted that there is a flat hydrophobic surface of at
least 7 ⫻ 12 Å extending from the sugar binding site with
an orientation similar to the plane of the pyranose. Many
of these inhibitors are relatively flexible, and conformational analyses predict folded solution structures. A study
focused on phlorizin (242) predicted the binding structure
being a folded conformation occupying a 13 ⫻ 10 ⫻ 17 Å
volume, an arrangement similar to the structure shown by
Panayotova-Heiermann (161) and Figure 19. Independent
of the actual conformation, the pharmacophore predicts
that there are three locations where H-bonds can be donated and accepted, one point where an H-bond is donated from the protein, and one area in which an H-bond
is detrimental (66). Phlorizin can be docked into homology models of SGLT1 and -2 with glucose in the sugar
binding site and the aglycone in either the enol or keto
conformation (V. Chaptel and B. A. Hirayama, unpublished data).
Given that hSGLT2 is 59% identical to hSGLT1 and is
expected to share the same architecture, how, then, does
one produce a specific SGLT2 inhibitor?
Over the past decade, several pharmaceutical companies have attacked this problem experimentally by
modifying the phlorizin structure to enhance selectivity
for SGLT2 over SGLT1. Phlorizin has a greater than fourfold higher affinity for hSGLT2 (Ki ⬃40 nM) than hSGLT1
(⬃200 nM) (81, 88, 159). The first report of an effective
91 • APRIL 2011 •
www.prv.org
756
TABLE
WRIGHT, LOO, AND HIRAYAMA
5. Glucoside interactions with SGLT2
Substrate/Inhibitor
Transported glucosides (K0.5, mM)
2-Naphthyl-␤-D-glucose
Indican
Phenyl-␤-D-glucose
Arbutin
Helicin
Glufosfamide
Glucoside inhibitors (Ki, Kd, or EC50,
mM)
Phlorizin
Phloretin
4-Methylumbelliferyl-␤-D-glucopyranoside
8-Hydroxyquinoline-␤-D-glucopyranoside
Esculin
1-Naphthyl-␤-D-galactopyranoside
2-naphthyl-␤-D-galactopyranoside
Deoxyrhapontin
Rhapontin
5Br-6Cl-3-indolyl-␤-D-galactose
5Br-3-indolyl-␤-D-galactose
T-1095a
Sergiflozin
Compound 5
Dapagliflozin
6=-O-Spiroglucoside
6=-O-Spiroglucoside-Cl
2=-O-Spiroglucoside
Salicin
Helicin
2-Nitrophenyl-␤-D-glucopyranose
4-Nitrophenyl-␤-D-glucopyranose
hSGLT1
hSGLT2
0.54
0.054
1.64
1.44,14
1414
Poor16
0.118
0.318
pSGLT3
0.24
0.94
7.94
1.24
116
0.00026,8,9,11,18
0.05-0.146,11
84,6
64,6
154,6
0.94,6
74,6
0.0256
0.26
0.146
0.186
0.00028,9
0.002111
⬎0.0088
0.00148
0.01-0.110
0.0006210
⬎.000110
0.000071,8,9,11,18
0.02511
0.00913
3.34
⬃404
NI4
NI4
NI4
0.000038,9
0.000018,11
0.00138
0.0000018
0.00007110
0.00000710
⬎⬎0.0000110
NI13
NI13
NI13
transported 33% ␣MDG13
Glycosides can be transported or be inhibitors. The apparent affinities of these compounds for hSGLT1, hSGLT2, and pSGLT3 are compared as
K0.5 for transported substrates, or Ki or EC50 for inhibitors. *Ki, or estimate of inhibition. ¶Value estimated from graph in paper. NI, no interaction.
§EC50. ‡Kd. #rbSGLT1. References are as follows: hSGLT4: 1Ref. 214; hSGLT1: 4Ref. 37, 5Ref. 38, 6Ref. 66, 17Ref. 67,18Ref. 81, 19Hirayama and Wright,
unpublished data; rat SMIT2: 2Ref. 3,15Ref. 21; canine SMIT1: 3Ref. 58; hSGLT3: 7Ref. 235,15Ref. 35, 13Ref. 138; hSGLT2: 8Ref. 81, 9Ref. 156, 10Ref. 255,
11
Ref. 159,12Ref. 86; rbSGLT1: 14Ref. 136; pSGLT3: 16Ref. 234.
SGLT2 inhibitor was T-1095A, a phlorizin derivative created by research groups at Tanabe Seiyaku (156), that
could be delivered orally in a pro-drug strategy. Selectivity for SGLT2 was accomplished by substituting a methyl
group for the meta hydroxyl of the B-ring, and creating an
aromatic five-membered ring of the para-OH of the A-ring
(Fig. 18A). The pro-drug was absorbed without interacting with the intestinal SGLT1, and the active drug had a
fourfold increase in SGLT2/SGLT1 selectivity.
We will present a brief synopsis of the continuing
evolution of specific hSGLT2 inhibitors. A more detailed
discussion of intermediate structures isavailable in pharmaceutical reviews (e.g., Refs. 83, 147, 237). Since the
description of T-1095A, several groups have addressed the
SGLT1/SGLT2 specificity problem. Approaches of note
are Sergiflozin-A, Dapagliflozin, and 6=-O-spiro-C-aryl glucosides (Fig. 19), as each represents advances in the
phlorizin modification strategy. In the initial investigation
of glucoside transport and binding to SGLT1, the importance of the glycosidic oxygen was noted, as was the
observation that a ␤-linked hydrophobic/aromatic structure could be accommodated in the binding site. This
Physiol Rev • VOL
model is followed in phlorizin, T-1095A, and Sergiflozin-A.
Sergiflozin-A is a ␤-glucopyranoside in which the A-ring is
joined to the B-ring by only one carbon rather than the
three carbons of phlorizin and T-1095A (88). Both hydroxyls of the B-ring and the carbonyl oxygen in the linker
are absent. The para hydroxyl of the A-ring of phlorizin is
replaced by a methoxy group. These changes resulted in
an increase in SGLT2/SGLT1 selectivity (to 210:1) and an
increase in SGLT2 affinity (Ki 2–10 nM) (88, 159).
Dapagliflozin is a further departure from the phlorizin plan (147) as it is a C-aryl glucopyranoside, i.e., the
B-ring is directly linked to the pyranose ring in an equatorial configuration, eliminating oxygen 1. The A- and
B-rings are still linked by a single carbon, but the B-ring
has a chloro group para to the sugar, and the A-ring is
attached in the meta position. The A-ring has a para
ethoxy group. These modifications improved SGLT2:
SGLT1 selectivity (1,200:1) and affinity for SGLT2 (EC50 ⫽
1.1 nM). Given the specificity profile of hSGLT1 (Table 5),
this increase in affinity in the absence of the glycosidic
oxygen was initially surprising, but we note that this
requirement may not hold for all SGLT isoforms (Table 2):
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
757
Ki (or rather EC50) values for these inhibitors increased
by roughly an order magnitude (Fig. 18). The pharmaceutical implications of the SGLT2 inhibitors are addressed in
section XIVH.
IX. ION SELECTIVITY
FIG. 18. Phlorizin and a 3-dimensional pharmacophore for hSGLT1.
A: chemical structure of phlorizin showing the A and B rings. B: an
inhibition pharmacophore for hSGLT1 was constructed by comparing
strength of inhibition of hSGLT1 function by aromatic glucosides with
potentially interacting groups placed at different positions on the rings.
Red spheres indicate a region where hydrogen bond donation and
acceptance are favorable; the blue sphere is the location where only
hydrogen bond donation is allowed, and the green sphere designates
the area where hydrogen bonding is detrimental. The yellow region
shows the extent of the planar surface for aromatic/hydrophobic
interactions. The space was generated by overlaying seven structures
that represent the diverse regions and types of interactions tested:
salicin, arbutin, two conformations of 2-naphthyl galactopyranoside,
1-naphthyl galactopyranoside, rhapontin, and 5-Br-6Cl-3-indolyl-galactopyranoside. [From Hirayama et al. (66).]
in hSGLT3, the K0.5 for 1-deoxyglucose is only reduced by
a factor of 2 compared with the 20-fold decrease for
SGLT1.
The third example is an O-spiro modification of
dapagliflozin where the orientation of the B-ring is
locked in position by creating a five-membered ring of
the ␣-oxygen and the B-ring of the aglycone (Fig. 19)
(255). This compound showed that the presence of the
substituent at the para position of the A-ring was important for SGLT2 binding, and the 6=-O-spiro compound was preferred over the 2=-O-spiro construct,
helping to more clearly define the orientation of the
B-ring. However, no gain in selectivity over that of
dapagliflozin was reported.
The development of selective SGLT2 inhibitors was
mostly accomplished with an increase in affinity for
hSGLT2 and a reduction in affinity for SGLT1: the hSGLT1
Physiol Rev • VOL
SGLTs are exquisitely selective for Na⫹ as the energizing cation. Apart from H⫹ and Li⫹, no other monovalent cation can replace Na⫹ to drive glucose transport (82,
166). The cation K0.5 values were 4 mM for Na⫹, 12 mM
for Li⫹, and 7 ␮M for H⫹ at ⫺150 mV (68, 175). The
maximum velocities were similar, and Hill coefficients
were greater than 1. There was a kinetic penalty for
substituting Na⫹ with Li⫹ or H⫹ in that the apparent
affinity for sugar decreased by 1–2 orders of magnitude:
K␣0.5MDG increased from 0.15 to 4 and 20 mM. One interpretation of the cation selectivity data is that cation
binding initiates a change of the conformation of the
sugar binding site: the conformations induced by Li⫹
and H⫹ binding are perturbed relative to the Na⫹-bound
transporter. The agreement between the Na⫹ to glucose
transport stoichiometry (2:1) and the Na⫹ Hill coefficient in SGLT1 and pig SGLT3 has led to the conclusion
that these proteins have two strongly interacting Na⫹
binding sites (34, 139). On the other hand, hSGLT2 only
has one Na⫹ binding site (81). One study examined the
FIG. 19. SGLT2 inhibitors. Representative conformations of the
hSGLT2-selective inhibitors sergiflozin, dapagliflozin, and the 6-spiro
modification of dapagliflozin are compared with phlorizin. The structures are aligned on the glucopyranose. EC50 values (nM) for each
compound are listed for hSGLT1 and hSGLT2 for comparison. These
structures represent favorable (low energy) configurations resulting
from conformational searches (in vacuo) of the rotatable bonds.
91 • APRIL 2011 •
www.prv.org
758
WRIGHT, LOO, AND HIRAYAMA
effect of anions on SGLT1 activity (126) and showed
that replacing Cl⫺ with gluconate or MES [2-(N-morphoino)ethanesulfonic acid] decreased the apparent affinity of SGLT1 for Na⫹ (KNa
0.5 increased from 41 to 62
mM at ⫺50 mV), whereas the effect on apparent affinity
for ␣MDG was small (⬍10%). Cl⫺ had no effect on
maximal transport rate.
What does the crystal structure of vSGLT tell us
about the Na⫹ binding sites? First, it should be made clear
that crystallographic assignment of Na⫹ binding sites is
nontrivial, even for proteins with much higher resolution
structures than are currently available for membrane
transport proteins (157, 158). The problem is that Na⫹ has
a small ionic radius of 0.97 Å and the same number of
electrons as water, and there are no electron-dense cations that can substitute for Na⫹.
Even in the case of the highest resolution structure of
a cotransporter to date (LeuT 1.65 Å, Ref. 256), the two
Na⫹ binding sites (Na1 and Na2) were assigned based on
modeling density peaks as water or Na⫹, valence calculations, and consideration of coordination distances. [Significantly, neither Na1 nor Na2 sites have been identified
in Na⫹-independent transporters in the LeuT structural
family (50, 54).] Both LeuT Na1 and Na2 sites are near the
unwound sections of TM helices and involve five or six
coordinating atoms, mostly backbone carbonyls. The major difference between the two sites is that the negatively
charged group of the substrate (carboxyl group of leucine) contributes to the coordination of Na1, while Na2 is
constructed from polar/neutral groups (carboxyl oxygens
from Gly-20, Val-23, and Ala-351 and hydroxyl oxygens
from Thr-354 and Ser-355; see Fig. 16B). Molecular dynamic simulations of ion selectivity of the LeuT Na1 and
Na2 sites suggest, as predicted, the Na1 site is more
selective for Na⫹/K⫹ than Na2 (5,000 vs. 400) (152).
One sodium site corresponding to Na2 of LeuT has
been assigned to vSGLT (Fig. 16B) (1, 48). This is ⬃10 Å
away from the sugar binding site with the coordinating
residues from TM1 (A62, I65) and TM8 (A361, S364, S365).
An equivalent site is found in the Mhp1 and BetP transporters, but the Na1 site is only found in LeuT and BetP.
The presence of one or two sites is consistent with the 1:1
and 2:1 Na⫹/substrate coupling ratio reported for these
transporters. Experimental evidence for the importance
of the Na2 site S365A in vSGLT was the abolition of
transport by S365A (48), and in the homologous Na2 site
in hSGLT1 the mutation S392A reduces both Na⫹ and
sugar affinities (Loo, Hirayama, Sala-Rabanal, and Wright,
unpublished data). Molecular dynamic simulations of mutations at the LeuT Na2 site predict that the mutations
S355A and T354A increase the Na⫹/K⫹ selectivity and
T354A increase the Li⫹/Na⫹ selectivity (Sergei Noskov,
personal communication).
Close inspection of the superimposed Na2 sites in
core vSGLT, LeuT, and Mhp1 structures shows that TM1
Physiol Rev • VOL
and TM8 are shifted away from each other in vSGLT (1).
The Na⫹ coordinating distances are larger (3.1–3.8 Å) in
vSGLT than in LeuT (2.2–2.5 Å), suggesting that in vSGLT
the cation has access to the cytoplasmic aqueous vestibule. This is supported by molecular dynamic simulations
indicating that Na⫹ is bound tightly to the Na2 site in
LeuT, but in vSGLT Na⫹ can escape into the cytoplasmic
aqueous vestibule and, after a transient interaction with
D186, into the cytoplasm (117, 238).
X. KINETICS
The high level of expression of SGLT1 in heterologous systems has led us to revisit the questions of SGLT1
kinetics using voltage-clamp methods. One significant advance has been the ability to record the kinetics as a
function of membrane potential and external sugar, sodium, and phlorizin concentrations in a single cell. In
addition, it is now possible to control the composition of
the intracellular compartment using either the cut-open
oocyte or patch-clamp preparations (17, 47, 81, 191). Another major advance was our introduction of fast perturbation techniques developed for enzymology to the field
of cotransporters. We used rapid jumps in membrane
potential to record the transient capacitive SGLT currents
as a function of external Na⫹, sugar, and phlorizin (presteady-state kinetics)(127). The significance is that for the
first time one can study partial reactions in the overall
transport cycle.
Relatively little is known about the transport kinetics
of other human SGLTs: SGLT2 is expressed poorly in
heterologous expression systems, where the rate of transport is generally orders of magnitude less than that for
SGLT1 (81, 86). Human SGLT3 is not a transporter but
instead a glucose sensor (35). Pig SGLT3 (originally designated pig SGLT2) is a transporter with a common transport mechanism to hSGLT1 (34, 138). Only limited information is available for SGLT4, -5, and -6 (SMIT2) (21, 113,
118, 214).
A. Steady-State Kinetics
1. Inward transport
As discussed above (Figs. 3 and 4), the activity of
hSGLT1 can be monitored as an inward sodium current
generated by glucose, and the kinetics can be determined
by recording the currents as a function of voltage and the
external sugar, cations, and inhibitor concentrations (see
Fig. 4B for ␣MDG and 3FDOG in a single oocyte). The
protocol that is commonly employed in oocytes expressing hSGLT1 is illustrated in Figure 20. The cell membrane
potential is voltage clamped at ⫺50 mV and then rapidly
stepped to voltages ranging from ⫹50 to ⫺150 mV for 100
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
759
⫹
FIG. 20. Kinetics of Na /sugar transport by hSGLT1. Total current records from an oocyte expressing hSGLT1 bathed in NaCl buffer are shown.
A: in the absence of sugar. B: with 0.3 mM ␣MDG added to the external solution. Membrane potential was clamped at ⫺50 mV and stepped to a series
of test voltages for 100 ms (⫹50 to ⫺150 mV). C: current-voltage relations of the sugar-induced current, obtained by subtracting the baseline current
in the absence of sugar from the current in the present of sugar. D: dependence of the sugar-induced current on sugar concentration. The curve was
drawn according to the equation: I ⫽ Imax [␣MDG]/([␣MDG] ⫹ K␣0.5MDG). E: dependence of K␣0.5MDG on voltage. F: dependence of KNa
0.5 on voltage.on
voltage. (From B. Moy, C. Lu, D. D. F. Loo, and E. M. Wright, unpublished data.)
ms, first in the absence of substrate (Fig. 20A) and then in
the presence of different external cation and sugar concentrations (Fig. 20B). Note that the practical range of
test potentials, ⫹50 to ⫺150 mV, is limited by the stability
of the cell membrane, i.e., without artifacts from voltageactivated endogenous channels or dielectric breakdown.
In the experiment illustrated in Figure 20, the external Na⫹ concentration ([Na⫹]o) was 100 mM, and the
␣MDG concentration ([␣MDG]o) was varied between 0
and 10 mM. In both the presence and absence of sugar,
the currents approached a steady-state value 95–100 ms
after stepping the membrane potential (Vm) to each test
value, and returned to the original holding potential with
a similar time course. The difference between the steadystate currents obtained in the presence and absence of
sugar at each voltage step gives the current-voltage (I-V)
curve at each sugar concentration (Fig. 20C). Note that
the I-V curves approach zero current at positive Vm values
and saturate at hyperpolarizing voltages, i.e., transport
has voltage-sensitive (between ⫹50 and ⫺100 mV) and
voltage-insensitive (⫺100 to ⫺150 mV) kinetics.
The substrate-coupled current (I) was empirically
described by the relationship:
Physiol Rev • VOL
S
S n
[S]n ⁄ {(K0.5
) ⫹ [S]n}
I ⫽ Imax
(1)
where [S] is the external substrate concentration (cation
S
or sugar substrate), Imax
is the maximal substrate-induced
S
current, K0.5 is the half-maximal substrate concentration,
and n is the so-called Hill coefficient.
In the experiment illustrated, the K0.5 for ␣MDG was
0.7 mM at ⫺50 mV (Hill coefficient set to 1) (Fig. 20D),
and the K0.5 decreased towards a minimum value of 0.2
mM at ⫺150 mV (Fig. 20E). In the same oocyte, the
currents induced by ␣MDG were also recorded as the Na⫹
concentration was varied between 0 and 100 mM and fit to
Equation 1 (not shown), and the apparent K0.5 for Na⫹
was plotted against voltage in Figure 20F. The Na⫹ K0.5
was voltage dependent ranging from 60 mM at 0 mV to a
minimum value of 0.8 mM at ⫺150 mV. The Hill coefficient
for Na⫹ was 1.5 and independent of voltage.
Steady-state kinetics provides insights into the order of substrate binding, as well as whether the substrates are transported sequentially or simultaneously.
For example, in a strictly ordered system, if glucose is
the last molecule to bind, then the maximal transport
S
rate for sugar (Imax
, determined by maintaining [Na⫹]
91 • APRIL 2011 •
www.prv.org
760
WRIGHT, LOO, AND HIRAYAMA
constant while varying sugar concentration) is independent of Na⫹ concentration; conversely, if Na⫹ binds
Na
last, Imax for Na⫹ (Imax
, determined by maintaining
[glucose] constant while varying [Na⫹]) is independent
of glucose concentration.
We found that the Imax for sugar (␣MDG) at saturating voltages (⫺150 mV) was independent of the Na⫹
concentration (Fig. 21C), whereas Imax for Na⫹ depends
on the glucose concentration (Fig. 21D). These observations indicate that external Na⫹ bind to the transporter
before the external sugar (167). At saturating voltages,
⫺150 mV, the K0.5 for Na⫹ depends on the ␣MDG concentration (Fig. 21B), and conversely, the K0.5 for ␣MDG
depends on the Na⫹ concentration (Fig. 21A). This mutual
dependence of the K0.5 values for Na⫹ and glucose indicates that two substrates are transported simultaneously.
The Hill coefficients for ␣MDG and Na⫹ are 1 and ⬎1.5,
respectively, consistent with measurements of sugar and
Na⫹ stoichiometry. The binding of the two Na⫹ to the
transporter before glucose is also supported by fluorescence experiments (see Fig. 25B).
Na⫹/glucose cotransport has large temperature dependence, temperature coefficient Q10 ⬎3, or activation
energy Ea ⬎20 kcal/mol, i.e., much greater than for chan-
nels (145, 168). There was no temperature effect on the
apparent affinities of rabbit SGLT1 for either Na⫹ or
glucose, but the maximum rate of transport increased
with a Q10 of ⬃3 (168).
2. Outward transport
Outward Na⫹/sugar currents were also measured
as a function of voltage and internal Na⫹ (0 –500 mM)
and sugar (0 –500 mM) concentrations when the external solution contained 10 mM Na⫹ and no sugar, i.e.,
the reverse of inward current measurements (see Fig. 6;
Refs. 47, 191). In contrast to the inward Na⫹/sugar
currents, which saturated with large hyperpolarizing
voltages and approach zero at large depolarizing voltages (Fig. 20C), the outward Na⫹/sugar currents saturated at 0 mV and approached zero at ⫺150 mV. The
internal K0.5 for sugar was ⬃40 mM at 0 mV (with
internal Na⫹ concentration of 500 mM). The outward
current decreased as the internal Na⫹ was lowered
from 500 to 0 mM with a sodium K0.5 of 45–50 mM when
the data were fit with a Hill coefficient of 2. Model
simulations indicate that if sugar is dissociated before
Na⫹, there would be sugar trans-inhibition, i.e., inhibi-
FIG. 21. Dependence of the steady-state kinetic
parameters K0.5 and Imax on [Na⫹] and [␣MDG]. The
experiment was performed on an oocyte expressing
the hSGLT1. A and C show the dependence of K0.5 and
Imax for ␣MDG on [Na⫹]. The K0.5 values are shown for
Vm at ⫺70 and ⫺150 mV. The maximal transport rate
(Imax) was determined at ⫺150 mV. B and D show the
dependence of the K0.5 and Imax for Na⫹ on [␣MDG].
Error bars represent the SE of the estimates. The
curves in A, B, and D were drawn by eye, and the
dashed line in C is the mean of the data values. Same
cell as in Figure 20.
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
tion of inward Na⫹/glucose cotransport at hyperpolarizing membrane potentials by internal sugar but not
Na⫹ (167). Sugar trans-inhibition has also been demonstrated in brush-border membrane vesicles (91).
B. Pre-Steady-State Kinetics
Understanding the mechanism of SGLT1 has been
revolutionized by our discovery of SGLT capacitive
currents, or charge movements. These are the carrier
counterparts of the gating currents of voltage-gated ion
channels and are thought to arise from movement of
charged and/or polar residues in response to changes in
the membrane electric field (11, 127–129, 166; see also
Refs. 18, 59, 99, 123). The carrier-mediated capacitive
transients are elicited by step jumps in membrane voltage in the absence of glucose (Fig. 20A). These transients, or pre-steady-state currents, are blocked by saturating concentrations of glucose and phlorizin (Figs.
20B and 22, B and C). Pre-steady-state currents are a
general property of cotransporters (see Ref. 59). So far,
we have been unable to detect any hSGLT2 pre-steadystate currents (81).
The pre-steady-state current of hSGLT1 is illustrated in
Figure 20A. In the absence of external sugar, the membrane
potential was held at ⫺50 mV (Vh) and then stepped to a
761
series of test values (from ⫹50 to ⫺150 mV in 20-mV decrements) for 100 ms (ON) before returning to Vh (OFF).
In response to the voltage jump, the total current consisted of 1) an initial membrane bilayer capacitance
transient which decays to steady state with a time
constant ␶ of 1 ms, which is insensitive to Na⫹ and
sugar concentrations. This component was also observed in noninjected control oocytes. Integration of
this transient in control oocytes gave a capacitance of
275–340 nF, and, with the assumption of a specific
membrane capacitance of 1 ␮F/cm2, the area of the
oocyte cell membrane area was 30 ⫻ 106 ␮m2 (262). 2)
SGLT1 pre-steady-state currents with time constants
greater than for the membrane capacitance ␶ vary between 3–20 ms, and 3) SGLT1 steady-state currents
consist of the background (endogenous) currents of the
cell and the Na⫹ currents mediated by SGLT1 (leak or
uniporter currents) (166, 131). During the ON and OFF
pulses, the SGLT1 pre-steady-state currents were in the
opposite direction, but the total charge moved (Qmax)
for ON and OFF voltage jumps were identical. The
SGLT1 current transients were inhibited by glucose
(Fig. 20B) and phlorizin (Fig. 22, B and C) and were not
observed in control cells. Increasing external ␣MDG
concentrations reduced the charge movements (Qmax)
with an apparent K0.5 of 1.1 mM, and phlorizin inhibited
FIG. 22. Dependence of hSGLT1 presteady currents on phlorizin and glucose. A: compensated current records (at Vm ⫹50, ⫺10, ⫺90, and ⫺130 mV) in
100 mM NaCl buffer. Vh ⫽ ⫺50 mV. B: current records
with 100 nM phlorizin added to the external solution.
C: dependence of charge movements (Q) on phlorizin.
The curves were obtained from the Boltzmann relation
and have been shifted to align at the depolarizing limit
(in the absence of phlorizin). Maximal charge Qmax is
the difference in Q between the depolarizing and hyperpolarizing limits. Qmax is decreased with increasing
[phlorizin]. V0.5 is the midpoint of the Q-V curve and is
unaffected by [phlorizin]. For this oocyte, V0.5 is ⫺37 mV.
D: relation between Qmax and maximal sugar transport
rates Imax. Slope of the regression line is 84 ⫾ 7 s⫺1.
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
762
WRIGHT, LOO, AND HIRAYAMA
with a Ki of 100 nM (Fig. 22C). The reduction in charge
by sugar was directly proportional to the increase in
sugar current with a slope of 84 s⫺1 (see Fig. 22D).
The pre-steady-state current of SGLT1 can be isolated from the total current (Itot) by two equivalent
methods (18, 59, 123, 127): 1) point to point subtraction
of the total currents in the presence and absence of
saturating phlorizin and 2) fitting the total current by
the equation
Itot(t) ⫽ Icm exp(⫺ t ⁄ ␶cm) ⫹ Ipss exp(⫺ t ⁄ ␶pss) ⫹ Iss
(2)
where Iss is the steady-state current, Icm exp(⫺t/␶cm) is
the bilayer capacitance current with initial value Icm and
time constant ␶cm, and Ipss exp(⫺t/␶pss) is the SGLT1
pre-steady-state current with initial value Ipss and time
constant ␶pss.
SGLT1 pre-steady-state currents were obtained from
Itot(t) by subtraction of the capacitive and steady-state
components [Ipss(t) ⫽ Itot(t) ⫺ Icmexp(⫺t/␶cm) ⫺ Iss]. The
“compensated current” records, i.e., the currents corrected for the bilayer membrane capacitive component
and the steady-state current, are shown in Figure 22, A
and B. For jumps to large depolarizing voltages, e.g., ⫹50
mV, the current rose to a peak (at 1.5 ms from the onset
of the voltage pulse) before decaying to the steady state
(128). The implications of the rising phase of charge
movement are discussed below (Fig. 25). The total charge
transfer Q at each test voltage (Vm) was obtained by
taking the integral of the current transients, and this was
equal for the ON and OFF responses. Q shows a sigmoidal
dependence on Vm (Figs. 21C and 22D), and the charge
versus voltage (Q-V) curve is empirically fitted by the
Boltzmann relation
{(Qdep ⫺ Qhyp) ⁄ Qmax ⫽ 1 ⁄ [1 ⫹ exp(z␦(Vm ⫺
V0.5)F ⁄ RT)}] (3)
where Qmax is the maximal charge transfer; Qmax ⫽ Qdep ⫺
Qhyp, with Qdep and Qhyp representing the charge measured at large depolarizing and hyperpolarizing limits,
respectively; and z␦ is maximum steepness factor for the
dependence of Q on voltage, and is the product of the
apparent valence of the movable charge (z) and the fractional distance (␦) within the membrane electric field in
which the charge moves. V0.5 is the midpoint voltage or
voltage at 50% Qmax (Fig. 22C). F is Faraday’s constant, R
is the gas constant, and T is absolute temperature. For
Qhyp to be the hyperpolarizing limit, we assume z ⬍ 0. In
this simple treatment the Boltzmann relation is interpreted as a distribution of a movable charge between two
states according to membrane voltage.
Pre-steady-state currents depend on external [Na⫹],
suggesting that Na⫹ can bind to the transporter in the
Physiol Rev • VOL
absence of the substrate. This is illustrated in Figure 23,
A–C, where [Na⫹]o was decreased from 100 to 25 mM.
Lowering the external [Na⫹] causes a decrease of the
transient currents in the depolarizing direction and an
increase in the hyperpolarizing direction. Compared with
100 mM Na⫹, the midpoint of the distribution, V0.5, shifts
to more negative values. The Q-V curves at 100, 50, and 25
mM Na⫹ were shifted to align at the depolarizing limit
(Fig. 23D), and the plots show that reducing external
[Na⫹] reduces the maximal charge (the difference between the hyperpolarizing and depolarizing limits) as well
as shifted the midpoint voltage V0.5. For the wild-type
hSGLT, it is not possible to obtain the Qmax at 0 mM Na⫹
as the Q-V curve moves out of the working range of the
test voltages. However, with several hSGLT1 mutants,
e.g., Y290C, substantial charge movements were observed
in Na⫹-free solutions, and the complete Q-V curve fell
within the ⫺150 to ⫹50 mV range. This demonstrated that
orientation of the Na⫹-free protein contributed to the
SGLT1 capacitive charge.
Attempts to identity the origin of SGLT1 charge
movement by mutation of charged residues in the membrane domain have not yet been successful, but this
should be revisited in light of the crystal structure.
Given that the Na2 binding site in vSGLT is a neutral
site (see above), we have to entertain the possibility
that protein dipoles contribute to the charge movements.
The dependence of the V0.5 on Na⫹ is shown in
Figure 23E. The line was obtained by linear regression
with slope 93 mV/10-fold change in [Na]o. The shift in V0.5
with [Na⫹]o suggests that the distribution of conformations of hSGLT1 in the membrane is dependent on [Na⫹]o.
The occupancy of SGLT1 proteins in the charge-generating conformations is increased with increasing [Na⫹]o
(see below).
Figure 23F shows the dependence of the relaxation
time constant (␶) of the pre-steady-state currents on
membrane voltage. For the ON pulse (filled symbols), in
the hyperpolarizing direction, ␶ (⬃20 ms) was relatively
independent of voltage, whereas ␶ decreased at more
positive test voltage to 3 ms (at ⫹50 mV). The time
constant for the OFF transients (open symbols), when
the membrane was stepped from the test voltage (Vm)
back to Vh, was independent of the test voltage (15 ms).
The time constants decreased as Na⫹ concentration
was reduced from 100 to 25 mM [Na]o (at 0 Na⫹ the
time constants are below the resolution of the 2-electrode voltage clamp). Using the cut-open oocyte preparation (see Fig. 26), pre-steady current in the absence
of Na⫹ decayed to steady state with a voltage-independent time constant of 0.24 ms for ON and 0.17 ms for
OFF voltage pulses.
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
763
⫹
FIG. 23. Dependence of hSGLT1 pre-steady-state currents on Na . A: total current records from an oocyte injected with cRNA coding for
hSGLT1. External medium was NaCl buffer. Membrane potential was held at Vh ⫽ ⫺50 mV and stepped (ON) to various test voltages (Vm) between
⫹50 and ⫺150 mV (in 20-mV decrements) for 100 ms before returning to Vh (OFF). B: compensated (or pre-steady state) current records at Vm ⫹50,
⫺10, ⫺90, and ⫺150 mV. The currents were obtained from the total currents by subtraction of the membrane capacitive transient and steady-state
current. C: current records at 25 mM [Na]o. D: charge transfer. Q was obtained from integration of the pre-steady-state currents. Q was the same
for ON and OFF responses, and shown are the QOFF. At each voltage, the Q-V data were fitted (smooth curves) to the Boltzmann relation (Eq. 3).
The Q-V curves at 100, 50, and 25 mM Na⫹ were shifted to align at the depolarizing limit. E: dependence of V0.5 on Na⫹. The slope of the regression
line is 93 mV/log [Na]. F: dependence of relaxation time constant (␶) on test voltage Vm. Shown are the ␶ obtained at 100, 25, and 0 mM [Na]o. Na⫹
free was obtained with choline replacement. Solid symbols correspond to ON pulses, and open symbols are the OFF. The time constants for OFF
pulses (from Vm back to Vh ⫽ ⫺50 mV) were independent of Vm.
1. Effect of sugar
The effect of sugar on pre-steady-state current is
illustrated in Figure 20B for an oocyte in 0 and 0.25 mM
␣MDG. The most pronounced effect of sugar was the
reduction in pre-steady-state current with hyperpolarizing
voltages. With increasing [␣MDG] (from 0 to 100 mM),
there was a reduction and eventual elimination of the
maximal charge Qmax. The reduction in Qmax was hyperbolic with a ␣MDG K0.5 of 1.1 mM. The midpoint voltage
(V0.5) shifted to more positive values with K0.5 values
similar to the K0.5 for reduction of Qmax with [␣MDG], but
z␦ was unaffected (127, 130). The reduction of Qmax and
shift of V0.5 with increasing sugar concentrations demonstrated that under sugar transporting conditions, there is
a partition of SGLT1 into non-voltage-dependent states.
Since Qmax provides an index of the number of transporters in the oocyte plasma membrane, the turnover rate
of SGLT1 can be estimated from the ratio (Imax/Qmax) of
the maximal rate of transport (Imax) and maximal charge
(Qmax), assuming z ⫽ 1. The turnover rate for ␣MDG
transport by hSGLT1 at 20°C was initially estimated to be
Physiol Rev • VOL
57 s⫺1 (127). It was revised to 28 s⫺1 when longer duration
voltage pulses (500 ms) revealed a slow component of
charge movement (time constant ␶ ⬃100 ms) contributing
a maximal charge similar to that of the 20 –30 ms (medium) component (128).
2. Phlorizin
Phlorizin also blocks the pre-steady-state currents.
This is illustrated in Figure 22, which shows the presteady-state currents in absence of phlorizin (A) and with
100 ␮M phlorizin (B) in the bath solution. Analysis of the
Q-V relations as external [phlorizin] varied from 0 to 250
␮M indicates that the Qmax was reduced with a phlorizin
Ki of 100 nM (Fig. 22C), but z␦ and V0.5 were unaffected
(59, 66, 130). The high affinity of SGLT1 for phlorizin (Ki
200 nM) means that during the 100-ms pulse, the inhibitor
has locked the transporter in a phlorizin-bound state.
Increasing the duration of a depolarizing pulse up to 2 s
shows partial recovery of charge and so demonstrates the
slow inhibitor off rate (130, see also 81).
91 • APRIL 2011 •
www.prv.org
764
WRIGHT, LOO, AND HIRAYAMA
3. Correlation between charge and expression
Qmax values at saturating Na⫹ can be used to estimate the number of SGLT1 proteins (N) in the plasma
membrane, N ⫽ Qmax/ze, where z is the apparent valence
of the moveable charge (the limiting slope of the Q-V
curve) and e is the elementary charge (127). In an oocyte
with a Qmax of 25 nC, this corresponds to ⬃1011 SGLT1
molecules/cell or 104 ␮m⫺2.
Independent confirmation that the maximal charge
provides an index of the number of cotransporters in the
oocyte plasma membrane was obtained from freeze-fracture electron microscopy (262). Replicas of the P or cytoplasmic face of the plasma membrane of SGLT1-cRNAinjected oocytes showed an increase in the density of
intramembrane particles (IMPs) with diameter 7– 8 nm
compared with control oocytes (262). There was a direct
correlation between particle density and charge density,
and the slope gives an estimate of ⬃3.5 charges per
SGLT1 molecule. The apparent discrepancy between the
apparent valency (z␦ ⫽ 1) and this estimate of charge may
in part be due to our simplifying assumption that the
SGLT1 charge movements occur in a single step by fitting
the Q-V curves to the Boltzmann relation (262).
Subsequently, we were able to demonstrate that
these IMPs in SGLT1 expressing oocytes were indeed
SGLT1 monomers (46). We concluded that Qmax is a valid
measure of the number of cotransporters expressed in the
oocyte membrane.
4. Conformational changes
The demonstration that the charge movements are
directly associated with protein conformation charges
came from biochemical experiments where we measured
the accessibility of covalent probes to a cysteine residue
(Q457C) in the sugar binding site (38, 129). The hSGLT1
mutant Q457C has a similar apparent affinity (K0.5) for
Na⫹ as wild-type hSGLT1 and is able to transport sugar.
Sugar transport, but not the SGLT1 capacitive transients,
were abolished after alkylation of the mutant by methanethiosulfonate reagents (38, 129, 146). However, inhibition
of Cys-457 SGLT1 was dependent on the conformation of
the transporter, as determined by the rate of inhibition of the
sugar-induced current after exposure to MTSEA (2-aminoethyl methanethiosulfonate hydrobromide). Figure 24A
shows an example. Initially, ␣MDG generated a current of
180 nA. After exposure to MTSEA in Na⫹ buffer (for 80 s),
the sugar-induced current was reduced to 40 nA. MTSEA
inactivation was reversed by washing the oocyte in DTT
(dithiothreitol, 10 mM). In the second part of the experiment, the inhibition of transport by MTSEA was blocked
by the presence of sugar (Fig. 24B). It was further demonstrated that the accessibility to Cys-457 to MTSEA was
dependent on the presence of external Na⫹ and was
blocked by phlorizin: there was no reduction in ␣MDGPhysiol Rev • VOL
FIG. 24. Dependence of effect of MTSEA on protein conformation
in hSGLT1 mutant Q457C. The ability of 1 ␮M MTSEA to label the
cysteine at residue 457 was influenced by substrate binding and membrane voltage. The oocyte was voltage-clamped at ⫺50 mV in Na⫹
buffer, and bath composition was varied. Effects of MTSEA were measured as the current generated by 200 mM ␣MDG. Between all of the
experiments, the oocyte was washed in 10 mM DTT for 15 min before
equilibration in Na buffer. Dashed line indicates baseline holding current. A: control experiment demonstrating MTSEA inactivation of sugar
transport was reversible. B: sugar protected against MTSEA inactivation. C: dependence of MTSEA inactivation on Na⫹ and phlorizin.
D: correlation of the voltage dependence of Q457C accessibility and the
pre-steady-state charge movement. Accessibility was measured as the
extent of inhibition of the sugar-induced current (unity is 100%) after
1-min exposure to 10 ␮M MTSEA. The relative voltage sensitivity was
obtained by normalizing the accessibility at the holding potential (⫺50
mV) to the charge. At each test voltage, the charge Q was obtained by
integration of the pre-steady-state currents. The curve was the fit of the
Q-V data to the Boltzmann relation (Eq. 3). Qmax ⫽ 13 nC, z ⫽ 0.8 , and
V0.5 ⫽ ⫺21 mV. The data have been normalized between 0 and 1 (129).
induced current if the oocyte was pretreated with MTSEA
in the absence of Na⫹ (choline replacement) or in the
presence of Na⫹ and 100 ␮M phlorizin before testing for
␣MDG transport (Fig. 24B).
In addition to ligands, accessibility of Cys-457
hSGLT1 to MTSEA was dependent on membrane voltage. The accessibility of MTSEA to Cys-457 was determined when the membrane potential held at different
values between ⫺90 and ⫹30 mV, and is compared with
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
the Q-V relation of the pre-steady-state current in the
same oocyte (Fig. 24D). The close correlation between
charge movement and changes in MTSEA accessibility
demonstrated that the Q-V relations represent the distribution of SGLT1 proteins between the outward-facing Na⫹-bound conformation and the inward-facing
empty carrier.
5. Voltage-clamp fluorometry
The correlation between charge movement and protein conformational changes is further supported by voltage-clamp fluorometry (Fig. 25) measurements (129, 146).
Fluorophores such as tetramethyl-rhodamine are sensitive to their local environment. A change in fluorescence
therefore indicates that the protein conformation has
changed, for example, moving tetramethyl-rhodamine
765
from a hydrophobic to a polar environment (such as
aqueous external solution) would decrease its fluorescence intensity. The charge and fluorescence measurements are complementary. SGLT1 capacitive charges are
associated with global conformational changes of the
transporter [the apparent valence of the movable charge,
z ⫽ z1␦1 ⫹ z2␦2 ⫹ ѧ ⫹ zn␦n, is the sum of the product of
all movable charges (zi) and the displacement of each
charge (␦i) in the membrane electric field], while changes
of fluorescence reflect changes in the local environment
of the fluorophore.
In the absence of sugar, Na⫹ can interact with
SGLT1. This is illustrated in Figure 25A, which shows the
time course of rhodamine fluorescence on an oocyte expressing Q457C labeled with TMR6M (tetramethyl-rhodamine-6-maleimide) as the cation in the external solution is
⫹
FIG. 25. Dependence of fluorescence on Na . Oocytes expressing Q457C were labeled with TMR6M. The experiments were performed using
a two-electrode voltage clamp on the stage of an inverted epifluorescence microscope. Fluorescence intensity changes (⌬F) are quantified as percent
change from baseline (⌬F/F). ⌬F is expressed in arbitrary units (a.u.). 1 a.u. represents a ⌬F/Ftotal of ⬃1%. A: cation dependence of fluorescence.
Vm was ⫺50 mV, and the time course of ⌬F was monitored. External perfusing solution initially contained 0 Na⫹ (100 mM choline Cl). At the arrow,
the solution was changed to either 100 mM Na⫹, Li⫹, TEA⫹, or NMDG⫹. An upwards deflection of the traces indicates an increase in fluorescence
intensity. B: kinetics of Na⫹ activation. ⌬F was measured as a function of [Na⫹]. The ⌬F vs. [Na⫹] curves are sigmoid and fitted to Eq. 1: ⌬F ⫽ (⌬Fmax
[Na⫹]n)/{[Na⫹]n ⫹ (K0.5)n}, where ⌬Fmax is the maximal fluorescence for saturating [Na⫹], K0.5 is the half-maximal concentration, and n is the Hill
coefficient. At ⫺30 mV, ⌬Fmax ⫽ 7.4 a.u., K0.5 ⫽ 91 mM, and n ⫽ 2.0. At ⫺50 mV, ⌬Fmax ⫽ 7.5 a.u., K0.5 ⫽ 58 mM, and n ⫽ 1.9. At ⫺90 mV, ⌬Fmax ⫽ 8.2
a.u., K0.5 ⫽ 49 mM, and n ⫽ 1.8. C: time course of ⌬F with step jumps in voltage. Vm was initially at ⫺50 mV and then stepped to between ⫹90 and
⫺190 mV (with 20-mV increments). External solution contained 100 mM (top panel), 25 mM (middle panel), and 0 mM [Na⫹]o (bottom panel).
D: relationship between ⌬F and voltage. ⌬F is the difference in steady-state fluorescence between the test and holding voltages. The ⌬F vs. V
relations for 100 and 25 mM [Na⫹] were fitted to a Boltzmann relation (Eq. 3): (⌬F-⌬Fhyp)/⌬Fmax⫽ 1/[1 ⫹ exp(z(V ⫺ V0.5)F/RT)], where ⌬Fmax ⫽
⌬Fdep ⫺ ⌬Fhyp; ⌬Fdep and ⌬Fhyp are the ⌬F at depolarizing and hyperpolarizing limits; F, R, and T have their usual meanings; V0.5 is membrane
potential at 50% ⌬Fmax; and z is apparent valence of the charge sensor of the fluorophore. At 100 mM [Na⫹], z ⫽ 0.4, and V0.5 ⫽ ⫺50 mV. At 25 mM
[Na⫹], z ⫽ 0.4, and V0.5 ⫽ ⫺125 mV. For comparison, the curves have been normalized to ⌬Fmax observed in 100 mM Na⫹ and shifted to align at
the extrapolated depolarizing limit. E: dependence of V0.5 for fluorescence (⌬F) on Na⫹. The slope of the regression line is 98.5 mV/log [Na⫹]. [From
Meinild et al. (146).]
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
766
WRIGHT, LOO, AND HIRAYAMA
changed. External perfusing solution initially contained 0
Na⫹ (100 mM choline Cl), and membrane potential was
held at ⫺50 mV. At the arrow, the perfusing solution was
changed from choline buffer to either 100 mM NaCl, LiCl,
TEACl, or NMDGCl. On exposure to Na⫹, the rhodamine
fluorescence intensity increased to a new steady-state
value and returned to baseline (in choline) when Na⫹ was
removed from the perfusing solution (not shown). This
increase in fluorescence intensity was specific for Na⫹
and Li⫹, which can substitute weakly for Na⫹ in driving
sugar transport, whereas TEA⫹ and NMDG had no effect.
The kinetics of Na⫹ activation of hSGLT1 were determined by measuring the fluorescence increase (⌬F) as
external Na⫹ concentration varied from 0 to 100 mM with
the membrane potential held constant at ⫺30, ⫺50, and
⫺90 mV (Fig. 25B). At each holding potential, increasing
Na⫹ concentration increased fluorescence. Likewise, at
constant [Na⫹], hyperpolarizing membrane potentials increased fluorescence. The ⌬F versus [Na⫹] relationship
was sigmoidal, with a Hill coefficient of 2 at all membrane
potentials. We conclude in the absence of glucose, 2 Na⫹
bind to SGLT1 and that there is high cooperativity between Na⫹ binding to the two sites.
The time course of the fluorescence intensity changes
with step jumps of the membrane voltage is shown in Figure
24C. The experiments were performed with external solution containing 100 mM (top panel), 25 mM (middle panel),
and 0 mM Na⫹ (choline replacement, bottom panel). For
both hyperpolarizing and depolarizing voltage pulses, the
fluorescence signal reached a steady state and was maintained until the membrane potential was returned to the
holding value. From the holding potential, ⫺50 mV, depolarizing voltages decreased the fluorescence, whereas hyperpolarizing voltages increased fluorescence.
The relation between the change in fluorescence intensity (⌬F) and the test voltage (Vm) was sigmoidal, and
these were fitted to the Boltzmann equation (Fig. 24D)
with parameters ⌬Fmax (maximal fluorescence intensity
change), z␦ (apparent valence of the voltage sensor for
fluorescence), and V0.5 (voltage at 50% ⌬Fmax). There
was a reduction in ⌬Fmax with decreasing Na⫹ concentration. V0.5 shifted to more negative values with decreasing [Na⫹]o, and the shift was 98 mV/10-fold reduction in [Na⫹]o (Fig. 25E). The direction and amplitude
of the shift of the V0.5 for fluorescence, towards more
negative values with a sensitivity ⬃100 mV/10-fold decrease in [Na⫹]o, is similar to that for charge movement
(Fig. 23E). The apparent valence of the voltage sensor
for fluorescence (z␦) was 0.4 and was independent of
Na⫹ concentration. In Na⫹-free solutions, there was
also a change in fluorescence with voltage jumps. This
confirms that the empty transporter undergoes voltagedependent conformational changes between the external and internal membrane surfaces. The lower apparent valence observed for the fluorescence, 0.4, than for
Physiol Rev • VOL
charge, 1.0, indicates that the fluorophore reports on
local conformational changes.
6. Experimental limitations
The two-electrode voltage clamp, cut-open oocyte
voltage clamp, and patch-clamp techniques have been
used to study SGLT1 and other electrogenic cotransporters. Each has strengths and weaknesses. The advantage of
the two-electrode voltage clamp is that the oocyte is
stable for many hours, and the clamp settling time of
0.6 –1.0 ms (127) can be used to record electrical and
optical transients in the millisecond to second range. The
relatively slow clamp speed is due to the microelectrode
resistance (⬃0.5 M⍀) and the high membrane capacitance of the oocytes, ⬃300 nF. To measure kinetic events
in the microsecond range, the cut-open oocyte voltage
clamp with a settling time of 80 ␮s may be used (204, 208).
However, this system tends to be less stable than the
two-electrode voltage clamp. Patch-clamp techniques,
such as the excised giant patch, have also been employed
to study SGLT1, but here the disadvantage is membrane
instability and the difficulty of obtaining high electrical
resistance contacts (⬎109 ⍀) between the patch pipette
and the oocyte plasma membrane (47).
The concurrent employment of an independent optical method to monitor voltage-induced conformational
changes in SGLT1 overcomes some of the inherent limitations of charge measurements alone, since they are
subjected to different constraints than charge measurements. Charge is extracted from SGLT1 pre-steady-state
currents within a background of plasma membrane capacitive currents, SGLT1 ionic currents (uniporter currents), and membrane leakage currents, whereas SGLT1
optical records are measured against a stable background
fluorescence.
The human isoform was used to study SGLT1 charge
movement because the midpoint voltage of the distribution of the protein between the outside-facing and insidefacing conformations (V0.5) was close to the normal holding potential, ⫺50 mV. This means that over the practical
range of voltages that can be used with oocytes, ⫺150 to
⫹50 mV, the full charge movement can be recorded, but
this is not the case with other SGLTs with V0.5 values of 0
to ⫺10 mV, e.g., rabbit SGLT1 (59). This also is a problem
at low Na⫹ concentrations where V0.5 moves towards
more negative voltages (Fig. 23) and it becomes difficult
to estimate values of Qmax. Such limitations resulted in
early confusion about the origin of charge movements
where some concluded that they were wholly ion-well
effects.
7. Submillisecond charge movements
In two-electrode voltage-clamp experiments, we observed a fast (submillisecond) rising phase of charge
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
movement when membrane potential was stepped from
the holding to a large positive test voltage (e.g., from ⫺50
to ⫹50 mV), and simulations predicted a fast rising phase
of charge movement with depolarizing voltage pulses (59,
167). To study this fast charge, the cut-open oocyte voltageclamp technique (with a settling time of 80 ␮s) was combined with fluorescence measurements on the hSGLT1 mutant Q457C labeled by TMR6M (128). In 100 mM external
Na⫹ (Fig. 26A), depolarizing voltage steps evoked a
charge movement that rose initially to a peak (with time
constant 0.17 ms) before decaying to steady state (with
time constants, 2–30 ms and 25–100 ms) in the presence
of Na⫹ but not its absence. The time to peak (0.9 ms)
decreased with [Na⫹] (Fig. 26B). In absence of Na⫹,
charge movements decayed to steady state with three
time constants (0.2, 2, and 150 ms). Charge movement
was accompanied by fluorescence signals with similar
time courses (Fig. 26, C and D), indicating that the conformational changes monitored by charge movement are
reflected by local environment changes at or near Q457C.
In external Na⫹, a depolarizing voltage pulse is
thought to result in Na⫹ release from the Na⫹-bound
transporter and a subsequent reorientation of the empty
transporter from the outward-facing to the inward-facing
767
conformation. The development of the rising phase of
charge movement in SGLT1 (and Shaker K⫹ channels;
Ref. 8) is due to a decease in time constant with depolarization of the membrane and the contribution of charge
from secondary steps. (Fig. 23F), i.e., the major voltagedependent step of the Na⫹/glucose transport cycle is the
return of the empty carrier from inward to outward facing
conformations (128).
C. Conformational Dynamics
Simultaneous charge and fluorescence measurements can also be used to study the distribution of conformations of SGLT1 under sugar transporting conditions.
In these experiments, we took advantage of the hSGLT1
mutant G507C. The TMR6M-labeled mutant transporter
exhibited kinetics similar to that of wild-type hSGLT1,
before and after labeling of Cys507 by TMR6M (130).
Figure 27A shows a comparison of charge and fluorescence (⌬F) records in the absence and presence of saturating ␣MDG (100 mM). Addition of saturating sugar (100
mM ␣MDG) nearly eliminated the pre-steady-state current, but not the fluorescence signal (Fig. 27B). This indi-
FIG. 26. Correlation between fast charge and fluorescence. The experiment was performed using the cutopen oocyte voltage clamp on TMR6M-labeled Q457C.
The currents have been compensated for membrane capacitance and background current using the P/4 protocol
with a subtracting holding potential of ⫺150 mV (128).
External and guard solution contained 100 mM Na-methanesulfonate and internal solution contained 100 mM
K-methanesulfonate. Vh was ⫺80 mV, and test voltages
were ⫹50 and ⫺150 mV. A: pre-steady-state currents at
100 mM Na⫹. B: pre-steady-state currents at 0 Na⫹. C and
D: comparison of the rising phase of the pre-steady-state
current (I) and fluorescence (⌬F) in 100 mM [Na⫹]o (C)
and in absence of Na⫹ (D).
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
768
WRIGHT, LOO, AND HIRAYAMA
urating indican only reduced the hSGLT pre-steady-state
currents by 10% (Fig. 28, A and B), whereas ␣MDG eliminated them. This contrast in behavior of indican and
␣MDG indicated a difference in the rate-limiting step
between the two substrates.
Computer modeling (see below) indicated that the
rate-limiting step for indican transport is sugar translocation, whereas for ␣MDG, it is dissociation of Na⫹
from the internal binding sites. Because of the differences in turnover rates and the rate-limiting step between indican and ␣MDG, indican behaved as a competitive inhibitor of ␣MDG transport (Fig. 28D). In this
FIG. 27. Effect of sugar on charge and fluorescence. Comparison of
charge and fluorescence (⌬F) records in the absence and presence of
saturating ␣MDG (100 mM). Vh was ⫺50 mV, and test voltage varied
from ⫹50 to ⫺150 mV (20-mV decrements). Dashed lines are the zero
current or zero ⌬F levels. A: in NaCl buffer. B: in NaCl buffer with 100
mM ␣MDG. C: in the presence of saturating phlorizin.
cated the presence of an electroneutral conformational
step. In contrast, phlorizin blocked the pre-steady-state
current and the conformational changes (Fig. 27C).
D. Substrate and Drug Interactions
SGLT1, like many other cotransporters, has been
found to transport a wide variety of substrate analogs at
widely different turnover rates. In the glucose, nucleoside, and dipeptide cotransporters, drugs were transported at maximum rates ranging from 10 to 150% of that
for the natural substrate (37, 136, 188).
For example, indican (indoxyl-␤-D-glucopyranoside)
was transported by hSGLT1 at 10% of the maximal rate of
glucose but with a fourfold greater apparent affinity; K0.5
for indican is 80 ␮M versus 300 ␮M for glucose (37, 125). The
I-V curves for the indican and ␣MDG currents at saturating
concentrations of the sugars are shown in Figure 28C. SatPhysiol Rev • VOL
FIG. 28. Characteristics of indican currents from an oocyte expressing hSGLT1. Membrane potential was held at ⫺50 mV (Vh), and test
voltage pulses were applied (from ⫹50 mV to ⫺150 mV in 20-mV
decrements). Red traces represent total current, and black traces are the
records with the oocyte membrane capacitance subtracted. A: current
records in NaCl buffer alone. B: in presence of 600 ␮M indican. C: I-V
curves of the indican- and ␣MDG-induced steady-state currents.
D: inhibition of ␣MDG (0.25 mM) current by indican. Membrane potential was ⫺50 mV. The dashed line was drawn with a Ki of 210 ␮M,
obtained by simulation of the 8-state kinetic model (see text). Kinetic
parameters used were: k12 ⫽ 45,000 M⫺2s⫺1; k21 ⫽ 300 s⫺1, k1a ⫽ 600 s⫺1,
ka1 ⫽ 50 s⫺1, kab ⫽ 5 s⫺1, kba ⫽ 40 s⫺1, kb6 ⫽ 100 s⫺1, k6b ⫽ 100 s⫺1, k25 ⫽
0.01 s⫺1, k52 ⫽ 3.5 ⫻ 10⫺4 s⫺1, k56 ⫽ 5 s⫺1, k65 ⫽ 2,250 M⫺2s⫺1, k23 ⫽ 45,000
M⫺1s⫺1, k32 ⫽ 20 s⫺1, k45 ⫽ 800 s⫺1, k54 ⫽ 81,667 M⫺1s⫺1, k34 ⫽ 50 s⫺1,
k43 ⫽ 50 s⫺1, k27 ⫽ 250,000 M⫺1s⫺1, k72 ⫽ 12 s⫺1, k85 ⫽ 800 s⫺1, k58 ⫽
756,173 M⫺1s⫺1, k78 ⫽ 0.5 s⫺1, k87 ⫽ 0.5 s⫺1. Total number of transporters NT ⫽ 1.5 ⫻ 1012.
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
769
experiment, the ␣MDG current (generated by 0.25 mM
␣MDG) was inhibited by increasing concentrations of
indican, Ki of 247 ␮M. Simulations indicate that if
␣MDG is present at the K0.5 concentration, then the
indican Ki was twice the K0.5 for indican transport, i.e.,
there was classical competitive inhibition between indican and sugar. The competition between ␣MDG and
indican is due to the “trapping” of the transporters in
the indican-bound conformation (125).
In recent years, there has been increasing interest in
using transporters as vehicles for drug delivery. Our findings using indican and glucose indicate that competition
between substrates and drugs should be taken into consideration when targeting transporters as drug delivery
systems.
XI. KINETIC MODELING
A. Model
Many of the experimental observations on the kinetics of SGLT1 are accounted for by the six-state ordered
nonequilibrium kinetic model for Na⫹-dependent sugar
transport originally proposed by Parent et al. (167). The
model (Fig. 29) is based on the alternating access mechanism (see also Fig. 2 for an earlier 6-state equilibrium
model). In a transport cycle, accessibility to the ligand
(Na⫹ and glucose) binding sites alternates between external and internal membrane surfaces. The transporter has
six kinetic states, consisting of the empty transporter [C]
(states 1 and 6), the Na⫹ bound [CNa2] (states 2 and 5),
and the Na⫹- and sugar-bound transporter [CNa2S] (states
3 and 4) at the external and internal membrane surfaces
(Fig. 29). On the external membrane surface, two external
Na⫹ bind to the transporter before glucose, and glucose is
released at the internal membrane surface before the
Na⫹. The transition (C2 ↔ C5) is the Na⫹-uniport mode of
the transporter.
The empty transporter is assumed to have a valence
of ⫺2. Membrane voltage is assumed to affect Na⫹ binding/dissociation with the transporter (C1 ↔ C2) and
(C5 ↔ C6), and the conformational change of the empty
transporter across the membrane (C1 ↔ C6). Since the
Na⫹ bound protein [CNa2] is electroneutral, the sugar
binding (rate constants k23, k32, k45, k54) and translocation
steps (k34, k34) were assumed to be voltage independent.
B. Experimental Basis
The experimental basis for the kinetic model for
Na⫹/glucose cotransport is described above, namely, biochemical evidence for alternating access, indirect and
direct measurement of Na⫹:sugar stoichiometry, the kiPhysiol Rev • VOL
⫹
FIG. 29. Six-state ordered kinetic model for Na -glucose cotransporter. Scheme showing the six kinetic states and the rate constants for
the transitions between them. In the absence of ligands, the transporter
exists in two states (C1 and C6). At the external surface, two Na⫹ bind
to the transporter to form the complex C2Na2. The sugar-loaded transporter (C3Na2S) undergoes a conformational change (C3Na2S to
C4Na2S) resulting in Na⫹-glucose cotransport. The reaction from state
C2Na2 to C5Na2 (dashed line) is the uniport mode of Na⫹ transport by
SGLT1. Pre-steady-state currents are associated with the partial reactions C2Na2 ¡ C1 ¡ C6. Region I (gray) is the pre-steady-state currents.
II (red) includes external sugar binding/dissociation (C2Na2 ¡ C3Na2S)
and sugar translocation across the cell membrane (C3Na2S ¡ C4Na2S);
and III (blue) are the internal Na⫹ release (C5Na2 ¡ C6) and sugar
release (C4Na2S ¡ C5Na2) steps. The rate constants (kij, from state i to
state j) can be determined by isolating partial reactions in each region.
Region I describe the pre-steady-state currents (128). Rate constants in
region II can be estimated using indican transport (132). The rate
constants for internal reactions (III) can be estimated from the study of
reverse Na⫹-glucose cotransport (47). For the 6-state model, the rate
constants (estimated using partial reactions I, II, and III described
above) are as follows: k12 ⫽ 140,000 M⫺2s⫺1, k21 ⫽ 300 s⫺1, k16 ⫽ 600
s⫺1, k61 ⫽ 25 s⫺1, k23 ⫽ 45,000 M⫺1s⫺1, k32 ⫽ 20 s⫺1, k34 ⫽ 50 s⫺1, k43 ⫽
50 s⫺1, k45 ⫽ 800 s⫺1, k54 ⫽ 190,000 M⫺1s⫺1, k56 ⫽ 5 s⫺1, k65 ⫽ 2,250
M⫺2s⫺1, k25 ⫽ 0.01 s⫺1, k52 ⫽ 0.0005 s⫺1, k27 ⫽ 50,000 M⫺2s⫺1, k72 ⫽ 0.01
s⫺1, ␦ ⫽ 0.7, ␣= ⫽ 0. 3, ␣⬙⫽ 0. [Modified from Parent et al. (167).]
netics of forward and reverse Na⫹/glucose transport, substrate binding order, voltage dependence of external Na⫹
binding/dissociation and reorientation of the empty transporter, the number of transporters in the plasma membrane, and the cooperativity between the two Na⫹ binding
sites (47, 67, 68, 127–130, 139, 146, 166, 262). The goal of
a kinetic model is to summarize the experimental data in
a coherent conceptual framework and to use the model to
make predictions for designing experiments to challenge
the model.
The binding of two Na⫹ to the transporter was modeled to occur in one step (167). This simplification was
justified at high Na⫹ concentrations due to the strong
positive cooperativity between the Na⫹ sites (49). Na⫹
binding to SGLT1 exhibits high cooperativity (146). The
91 • APRIL 2011 •
www.prv.org
770
WRIGHT, LOO, AND HIRAYAMA
one-step approximation is an oversimplification, and it
fails to account for the experimental data as external
[Na⫹] is reduced towards zero (see below). However, key
properties of the kinetics of SGLT1 can be easier to see
with such a simple model.
C. Voltage Sensitivity
The effect of membrane voltage Vm on the rate constants was assumed to follow Eyring rate theory (with
symmetric potential energy profile between the forward
and backward reactions, Ref. 167). Dielectric parameters
␣ and ␦ are used to represent the fraction of the membrane electric field felt by Na⫹ binding and the translocation of the empty transporter (110), respectively. The rate
constants for a transporter with a valence (zT) of ⫺2 are
k16 ⫽ k016 exp(␦FVm/RT), where F is the Faraday constant, R is gas constant, and T is temperature. For external Na⫹ binding, k12 ⫽ k012 [Na]o2exp(⫺␣=FVm/RT) and
k21 ⫽ k021 exp(␣=FVm/RT); and internal Na⫹ binding: k65 ⫽
k065 [Na]i2exp(␣⬙FVm/RT) and k56 ⫽ k056 exp(⫺␣⬙FVm/
RT). The pre-steady-state currents (hSGLT1 capacitive
currents) were assumed to be associated with the reorientation of the empty transporter between outward- and
inward-facing conformations (C1 ↔ C6) and external (C2
↔ C1) and internal Na⫹ binding/dissociation (C5 ↔ C6).
The changes of fluorescence intensity associated
with SGLT1 with step jumps in membrane voltage were
assumed to arise from changes in occupancy probabilities: ⌬F ⬇ qy1⌬C1 ⫹ qy2⌬C2 ⫹ qy3⌬C3 ⫹ qy4⌬C4 ⫹
qy5⌬C5 ⫹ qy6⌬C6, where qyj is the apparent quantum
yield of the fluorophore (TMR6M) when SGLT1 is in conformation Cj (128).
D. Estimating Parameters
The Na⫹/glucose cotransport model involves a
closed cycle of six partial reactions, each of the phenomsug
Na
sug
enological kinetic parameters KNa
0.5, K0.5 , Imax, and Imax
depend on all the rate constants of the transport cycle
(see for example, Eqs. A37–A43 in Ref. 167). For the
six-state model, there are 14 rate constants, and it is a
challenge to estimate all the values. For their determination, we note that the partial reactions fall into three
groups: 1) voltage-dependent reactions [these are the conformational changes of the empty transporter between
external and internal membrane surfaces (C1 ↔ C6) and
external Na⫹ binding/dissociation (C1 ↔ C2)]; 2) external
sugar binding/dissociation (C2 ↔ C3) and sugar translocation (C3 ↔ C4); and 3) internal Na⫹ release (C5 ↔ C6)
and sugar release (C4 ↔ C5). Moreover, in two-electrode
voltage-clamp experiments where the inward transport of
Na⫹ and glucose are studied, the rate constants for sugar
Physiol Rev • VOL
translocation across the membrane (C3 ↔ C4) and the
internal ligand binding steps (C4 ↔ C5; C5 ↔ C6) cannot
be uniquely determined.
A set of rate constants for the six-state model for
hSGLT1 obtained from experiments where individual partial reactions were isolated is presented in Figure 29:
1) the kinetic parameters for voltage-dependent reactions
were obtained from fitting the pre-steady-state currents in
the absence of sugar; and 2) kinetic parameters for Na⫹
and glucose binding on the internal membrane surface
were obtained from studies on reverse sugar transport
using the excised giant patch (47). The rate constants for
sugar binding and translocation were estimated from forward Na⫹/glucose cotransport (68). These rate constants
were obtained from simulation of the global steady-state
and pre-steady-state SGLT1 kinetics, and not simply the
simulation of a single type of experiment.
There are several notable implications of the model:
1) the voltage-dependent steps are the reorientation of the
empty carrier and external Na⫹ binding/dissociation. The
internal Na⫹ binding steps do not depend on voltage.
Thus there is an asymmetry in the voltage dependence of
Na⫹ binding between the outside and inside; 2) the ratelimiting step of Na⫹/glucose transport is the release of
Na⫹ on the internal membrane surface; 3) under steadystate conditions, the current generated by Na⫹/glucose
cotransport arises from the “conformational” current of
the empty carrier as it returns from the internal to the
external membrane surface (C1 ↔ C6) and external Na⫹
binding/dissociation (C1 ↔ C2); and 4) the model accounts for the outward current, but this requires that the
rates of translocation of the empty transporter, k16 and
k61, are sodium dependent as predicted (49). The asymmetry in the I-V curves between inward and outward
currents are in agreement with our model where the
binding of Na⫹ at the internal surface is voltage independent. However, this is inconsistent with the location of the
Na⫹ binding site in the sugar occluded vSGLT crystal
structure (48).
E. Testing
Most of the tests of the six-state model have been
based on extending the type of experiment and the time
scale of the pre-steady-state currents, for example, examining the kinetics of different substrates (glucose and
indican) and reverse transport. The time scale of the
pre-steady-state kinetics was extended by 1) using a
faster voltage clamp, such as the cut-open oocyte voltageclamp method which has a 10-fold faster settling time (80
␮s compared with 1 ms) compared with the two-electrode
voltage clamp (18, 128); and 2) application of test voltage
pulses of duration longer than the 100 ms standard pulses
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
(100, 128). In the simplified three-state model for presteady-state currents (C2 ↔ C1 ↔ C6), the binding of the
two external Na⫹ is lumped into a single step. This is
essentially an equilibrium assumption at high Na⫹ conNa
Na
centrations, i.e., at [Na⫹]o ⬎⬎ KD
(KD
is the intrinsic
⫹
dissociation constant for Na binding to the protein). One
would expect model predictions to deviate from experimental data at low Na⫹ concentrations and during relaxations to steady state in response to step perturbations in
membrane voltage or ligand concentrations. Simulations
of the three-state model (59, 128) predict the following: 1) the
pre-steady-state current relaxation contains two time constants, a fast submillisecond component associated with
Na⫹ binding/dissociation and a medium component 2–30
ms associated with the empty transporter; 2) the dependence of the medium time constant (␶ON) on membrane
voltage is Gaussian or bell shaped with a peak at the V0.5,
and ␶ON is decreased with hyperpolarizing and depolarizing voltages away from the V0.5; 3) the shift of the V0.5 of
the charge movement is ⫺60 mV/10-fold decrease in
[Na⫹]o; and 4) since the empty transporter has a valence of ⫺2, z␦ (in the Boltzmann relation) should be 2.
While the steady-state kinetics are well described by
the six-state model, there are several predictions that
have not been met regarding the presteady state currents.
1) Charge and fluorescence measurements have revealed that the pre-steady-state current contains multiple
time constants with time scales ranging from 0.1 to 100
ms (128). In 100 mM external [Na⫹], depolarizing voltage
pulses evoked a charge movement that rose initially to a
peak (with time constant ␶ ⫽ 0.17 ms) before decaying to
steady state with two time constants (␶ ⫽ 2–30 and 25–150
ms). In the absence of Na⫹, charge movement decayed to
steady state with three time constants (0.2, 2, and 150 ms),
indicating that reorientation of the empty transporter between the outward (C1) and inward (C6) facing conformations involves at least two additional intermediate
states (C1 ↔ Ca ↔ Cb ↔ C6).
2) There is an asymmetry in the voltage dependence
of the time constants. The medium (␶med 2–30 ms) time
constant was independent of voltage for hyperpolarizing
pulses, whereas ␶med decreased with depolarizing voltage
pulses (18, 128). The asymmetry in voltage sensitivity has
also been observed in ion channels and has been attributed to an asymmetry in the potential energy profile (261).
3) The shift of the midpoint voltage (V0.5) with decreasing [Na⫹]o is ⫺100 mV/10-fold decrease in [Na⫹]o
(Fig. 23E). This is also the case for the shift of the V0.5 for
fluorescence with [Na⫹]o (Fig. 25E; Ref. 146). The failure
of the three-state model to account for the shift of the V0.5
with decreasing [Na⫹]o is not surprising, because lumping
the two Na⫹ binding steps into one step means that the
rate constants k16 for empty transporter reorientation
Physiol Rev • VOL
771
(C1 ↔ C6) and k12 for Na⫹ binding (C1 ↔ C2) become
pseudo rate constants and are dependent on [Na⫹]o (49).
Computer simulations of different models for Na⫹ binding to cotransporters showed that two-step sequential
models for Na⫹ binding are required to account for a shift
of 100 mV/10-fold change in [Na⫹] (Dr. Ian Forster, University of Zurich, personal communication).
4) z␦ determined experimentally is 1, whereas simulations predict a z␦ of 2. This discrepancy could, in part,
be due to our inability to measure very rapid (submillisecond) components of charge movement.
To account for three relaxation time constants in the
absence of Na⫹ and the asymmetry in voltage dependence, two additional states Ca and Cb between C1 and C6
(C1 ↔ Ca ↔ Cb ↔ C6) are required (region I of Fig. 29;
Ref. 128), together with asymmetry in the voltage dependence of the rate constants. Simulations showed the
steady-state kinetics of Na⫹/glucose cotransport was
equally accounted for by both the six- and eight-state
models. The models qualitatively and quantitatively accounts for I-V relations for forward and reverse transport,
as well as the dependence of the kinetic parameters KNa
0.5,
Na
␣MDG
K␣0.5MDG, Imax
, and Imax
on membrane voltage and ligand
concentrations (130). The models also account for indican transport (a high-affinity, low-turnover substrate) and
the competition between ␣MDG and indican (Fig. 28D).
F. Distribution of Conformations
Insight into the transport kinetics may be obtained by
an analysis of the occupancy probability (Po) of each state
in the eight-state model, and how those states are altered
by ligands and voltage. In external NaCl buffer (Fig. 30),
the transporter is predominantly in C2Na2 at negative
membrane voltages (98% occupancy in C2Na2 at ⫺150
mV), and in C6 at depolarizing voltages (73% at ⫹50 mV).
With step jumps in membrane voltage, the pre-steadystate current is associated with redistribution of the negatively charged transporter between the outward-facing
and inward-facing conformations (C2Na2 and C6).
External sugar alters the distribution by reducing the
occupancy in C2Na2 and increasing C5 occupancy at large
negative membrane voltages. At large positive voltages
(e.g., ⫹50 mV), the inward-facing ligand-free conformation C6 is predominant, regardless of the sugar concentration (Fig. 30, A, B, and D). At saturating external
[␣MDG], when membrane voltage is stepped from ⫺50 to
⫹50 mV, the charge movement is too small to be detected
because of the low occupancy in states C2Na2, C1, Ca, Cb,
and C6 (Fig. 30D). However, there is a conformational
change from C5Na2 ↔ C6, and is monitored by the fluorescence changes in the TMR6M labeled G507C (Fig.
27B). The fluorescence change is due to differences in
apparent quantum yields of the fluorophore between the
91 • APRIL 2011 •
www.prv.org
772
WRIGHT, LOO, AND HIRAYAMA
FIG. 30. Simulation of occupancy probabilities
(Po) of the 8-state model. For comparison of charge
and fluorescence measurements, the simulations
were performed for the TMR6M-labeled hSGLT1
mutant G507C, which exhibited similar steady-state
and pre-steady-state kinetics as wild-type hSGLT1.
For clarity, states with Po ⬍ 0.1 were not plotted.
These states were as follows: C1 in A, C, and E; C1,
C2Na2, C3Na2S, and C4Na2S in B and D. A: in NaCl
buffer in the absence of sugar. B and D: in 1 mM
and saturating (10 mM) ␣MDG. C and E: in 0.4 and
1 ␮M phlorizin. [From Loo et al. (125).]
two states, C5Na2 and C6. We predict that this conformational change is a rate-limiting step for Na⫹/sugar cotransport.
Phlorizin locked the transporters in a fixed state
(C7Na2Pz) and prevented the protein from running
through the transport cycle. The increase in the occupancy of the phlorizin-bound state (C7Na2Pz) is balanced
by reduction of the occupancy probabilities in all other
conformations (Fig. 30, A, C, and E). The profile of the
Po-Vm curves is not affected by phlorizin. At the Ki for
phlorizin, 50% of the transporters are in the phlorizinbound conformation. The effect of phlorizin on the presteady-state current is consistent with the model simulations: there is no effect of phlorizin on midpoint voltage
(V0.5), and maximal charge (Qmax) is reduced 50% at the Ki
for phlorizin.
The predicted occupancy in the different conformations with changes in Na⫹ and sugar concentrations and
Physiol Rev • VOL
membrane voltage (Fig. 30) was supported by the close
agreement between experimental data on fluorescence
and the simulations (125, 128, 130, 135).
Simulations using the eight-state model show that,
qualitatively and quantitatively, it accounts for the steadystate kinetics of Na⫹/glucose cotransport. However, it
fails to fit the pre-steady-state kinetics below 1 ms due to
assumptions about the kinetics of sodium binding (125,
135), e.g., 1) the initial rise component of the charge
movement with depolarizing voltage pulses (Fig. 26) (128)
and 2) magnitude of the shift of V0.5 (⬃100 mV/10-fold
change in [Na]o) for charge and fluorescence. These are
related: the rising phase is associated with the Na⫹ binding steps (128), and a sequential binding model with 2 Na⫹
binding sites (CNa2 ↔ CNa1 ↔ C1) is required to account
for the 100 mV (per 10-fold change in [Na]o) shift in V0.5.
At present, very little is known about the kinetics of the
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
two Na⫹ binding steps. A future challenge is to gain
insights into the binding of the Na⫹.
XII. STRUCTURE AND FUNCTION
The ultimate goal is to understand, on the atomic
level, how an ion-driven cotransport protein transforms
the energy of the electrochemical gradient into work,
enabling it to drive the cosubstrate against its concentration gradient. The kinetics studies have identified eight
different protein conformations, the rates of transitions
between them, and the dynamics of the conformational
changes in a Na⫹/sugar transport cycle (Fig. 31). With the
solution of two conformations of the SGLT1 bacterial
homolog vSGLT, providing for the first time a structural
model for SGLT1, the task ahead is to merge the kinetics
and structure into a dynamic structural translation. Even
though we do not yet have the structure of hSGLT1, the
ideal situation, there are tools available to begin the process.
At the time of this writing there are four different
Na⫹ cotransporter crystal structures in different conformations. These cotransporters, LeuT, vSGLT, BetP, and
Mhp1, belong to different gene families but share a common core architecture of 10 helices in an inverted repeat
configuration. We believe that this shared structural fold
implies a common kinetic mechanism; therefore, in the
following discussion we will use the LeuT helix numbering system. [For example, vSGLT TM1 is renamed TM-1
because vSGLT TM2 corresponds to TM1 of the core
773
structure (see Fig. 12), and so on.] The structure of the
nucleobase cotransporter Mhp1 has been solved in the
outward open (C2), outward occluded (C3), and inward
facing (C5) conformations (199). The leucine cotransporter LeuT is an outward occluded (C3) conformation
(256), BetP appears to be in an intermediate conformation
between C3 and C4 (183), and vSGLT1 is in an inward
occluded and inward open conformations (C4, C5) (48,
238). As yet, there are no structures of these proteins in
the C1 or C6 conformations.
Our first strategy was to “thread” the vSGLT1 sequence onto the LeuT coordinates to model the outward
occluded (C3) conformation (48). Comparison of the
vSGLT C4 configuration and the C3 model were then
compared to identify how the helices of the core have to
move in the C3 to C4 transition (alternating access).
There is an external aqueous pathway where extracellular sugar and Na⫹ gain access to the binding
sites (bounded by TM1E, TM2, TM6E and TM10), and
the aqueous pathway leading from the binding sites to
the cytoplasm (bounded by TM1I, TM2, TM3, TM6I,
TM8, and TM10). The major arrangements that are predicted to occur during the transition from the outward
(C3) to inward (C4) facing conformations involve TM1,
TM6, and TM10. The external end of TM10 folds over
the sugar binding site and F423 forms part of the external gate along with Y87 (TM2) and M73 (TM1) (see
Fig. 14C). The tilting of TM10 is facilitated by P436 and
G437 in the middle of the transmembrane domain.
Changes in the position of TM2, TM3, TM6I, TM8, and
FIG. 31. A 6-state model of SGLTs to
integrate the kinetic and structural data.
Na⫹ binds first to the outside to open the
outside gate (C2) permitting outside
sugar to bind and be trapped in the binding site (C3). This is followed by a conformational change from an outward occluded (C3) to an inward occluded (C4)
state. Upon opening the inward gate
(C5), the Na⫹ and sugar are released into
the cell interior. There is a paucity of
experiments adressing the order of the
ligand dissociation at the cytosolic surface. The transport cycle is completed by
the change in conformation from the inward facing ligand-free (C6) to the outward facing ligand free (C1) states. Structures corresponding to C2 and C3 have
been obtained for Mhp1, C3 for LeuT, and
C4 for vSGLT.
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
774
WRIGHT, LOO, AND HIRAYAMA
TM10 also result in the expansion of the intracellular
cavity. All that is then required to release sugar into the
expanded cavity is the displacement of the inner gate
Y263 (238).
Another view of the conformational changes that
follow external sugar binding was obtained by homology modeling of hSGLT1 based on the structures of the
Mhp1 obtained in the presence and absence of substrate
(241). Figure 32 shows an outside view of hSGLT1 before
and after sugar binding. In the outward facing open
conformation (C2), the hydrophobic gating residues
L87, F101, and F453 are open (12–20 Å apart), but these
gates close after glucose binding (⬍6 Å apart) to trap
the sugar. This occurs through a tilting of the external
end of TM10 into the vestibule, facilitated by the double
prolines in the middle of the helix (P465 and P466).
This proline duplex is well conserved in the SLC5 family (Table 1) and Mhp1 (GP), but not in the LeuT family.
In addition, the external ends of TM1 and TM2 tilt
inward to bring the sugar coordinating residues into
range. The unwound region in TM1 could facilitate the
inward motion in all four families of Na⫹ coupled transporters.
The predicted sugar-induced changes in tilt of the
external end of TM10 is supported by cysteine scanning
studies on vSGLT and hSGLT1 (67, 230). Specifically,
sodium-dependent glucose binding induced changes in
the fluorescence of probes covalently linked to cysteines
at the external end of TM10 (A423C in vSGLT, and Q457C
and D454C in hSGLT1). In addition, access of other mutants on the outer half of TM10, T460C and A468C, to MTS
FIG. 32. Model of the hSGLT1 structural changes at the sugar site
after glucose binding, and occlusion viewed from the external side of the
membrane. The open out, sugar-free structure is based on Mhp1, and the
closed out, sugar-occluded structure is based on vSGLT. Shown are TMs
1, 2, 6, and 10 along with the outer and inner gate residues [outer L87,
F101, and F453 (grey spheres), inner Y290 (yellow sphere), and sugar
coordinating (sticks)]. In the C2 conformation, the gates are open
(10 –20 Å apart) to expose the sugar binding residues. Upon sugar
binding, the external gates close (4 – 6 Å apart) largely through an inward
tilt of the external ends of TM10, TM1, and TM2. (From V. Chaptal, B. A.
Hirayama, D. D. F. Loo, J. Abramson, and E. M. Wright, unpublished
data.)
Physiol Rev • VOL
reagents was also conformationally dependent (36, 67).
Cysteine mutants lying outside the external vestibule also
exhibit conformational changes, either access to MTS
reagents (A439C, I443C, and Q445C at the top of TM9 and
the linking loop to TM10, R499C at top of TM11, and
L527C and Y528C at the top of TM12) (67). This indicates
that there are sugar-induced structural changes throughout the protein.
Open structural questions still to be resolved include
the following: 1) the identity of the second sodium binding site in hSGLT and determining whether it is a neutral
(Na2) or charged (Na1) site (152); 2) the location of the
inhibitor (phlorizin) binding site in hSGLT1 and hSGLT2.
Phlorizin-like inhibitors are in development for treatment
of diabetes (see Fig. 19). Phlorizin is a poor inhibitor of
vSGLT (Ki ⬃1 mM), but a potent inhibitor of human
SGLTs (Ki ⬍ 200 nM). 3) The structure of SGLTs in
additional conformations, C1/C6, C2, and/or C4/C5, need
to be determined.
XIII. MULTIFUNCTIONAL PROTEINS
SGLT1 does many things: it cotransports Na⫹ and
glucose, bringing an important nutrient into cells, but in
doing so it depolarizes the plasma membrane, which can
serve as a signal; in the absence of glucose, SGLT1 can
still support a Na⫹ current; the protein contains a water
and urea channel; and it transports water and urea along
with Na⫹ and glucose (249). It is highly likely that these
varied properties of SGLT1, and other cotransporters,
have yet undiscovered physiological significance. It is a
major challenge to identify the physiological functions of
cloned proteins such as those in the human SGLT gene
family. Even when genes have been cloned by expression
cloning, e.g., SGLT1, it remains a nontrivial task to establish their physiological role in humans. Furthermore,
there may be significant species differences, e.g., pig
SGLT3 behaves as a Na⫹/glucose cotransporter in the
Xenopus laevis oocyte expression system while human
SGLT3 does not (34, 35, 138). It appears that in humans,
SGLT3 is a glucose sensor expressed in the enteric nervous system and muscle. The first step in revealing the
diverse functions of these proteins is to determine the
properties of the cloned transporters in heterologous expression systems and to cast the net wide in terms of
functional assays. A second step is to carefully phenotype
patients with severe mutations in SGLT genes.
A. Naⴙ Uniport
The first indication that SGLT behaved as a uniporter
was the observation that phlorizin inhibited a current in
the absence of glucose (see Fig. 4A) (226). This SGLT
“leak” current was ⬃8% of the total Na⫹/glucose cotrans-
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
porter current and was not observed in control cells.
Further experiments (131, 162) demonstrated that this
current saturated with increasing external Na⫹ concentration with a K0.5 of 2.5 mM and a Hill coefficient of 2,
and that both Li⫹ and H⫹ were able to substitute for Na⫹.
The activation energy for the Na⫹ leak current was similar to that for Na⫹/glucose cotransport, 21 versus 25
kcal/mol. It was concluded that the leak currents reflect
Na⫹ uniport through SGLT1.
B. Water and Urea Channels
SGLT1 provides a passive permeability pathway for
water and other small molecules. Expressing SGLT1 in
Xenopus laevis oocytes increased the osmotic water permeability of the cell (262). Expressed in terms of the
number of SGLT1 proteins in the membrane, the osmotic
water permeability is similar to that for the water channel
AQP0, but only 5% of that for AQP1. The SGLT1 passive
water permeability was independent of the direction and
magnitude of the osmotic gradient, the presence or absence of ligands (Na⫹ and sugar), but in Na⫹ it was
blocked by phlorizin with a Ki of ⬃5 ␮M (124, 131, 134,
145, 262). Unlike the Na⫹ uniport mode, the activation
energy for water permeability is low (5 vs. 21 kcal/mol)
and similar to that for water channels. It is concluded that
SGLT1 behaves as a water channel, but unlike other water
channels, permeation is dependent on the protein conformation, i.e., channel activity is blocked by phlorizin.
Urea uptake into SGLT1 expressing oocytes was also
fourfold higher than in control oocytes (115, 165). The
SGLT1-specific urea transport was blocked by phlorizin
(Ki 1 ␮M), but only in Na⫹ buffer as phlorizin only binds
to SGLT1 in Na⫹. Phloretin also inhibits urea and water
transport through SGLT1, but the Ki was 100 –1,000
higher than that for phlorizin. As with SGLT1 water
permeability, the urea uptake was not affected by the
presence or absence of Na⫹, and the activation energy
was 6 kcal/mol. The activation energy for both water
and urea transport in control oocytes was 15 kcal/mol.
There was no saturation of SGLT1 urea transport as
uptake was not blocked by 100 mM concentrations of
urea analogs, thiourea, 1,3-dimethyl urea, 1,1-dimethyl
urea, or acetamide. It was concluded that SGLT1 behaves
as a urea channel. Examination of the crystal structure of
vSGLT and other members of the LeuT structural family
does not reveal an obvious continuous pathway for water
and urea across the protein, but the extracellular and
intracellular aqueous vestibules suggest that pathway is
through the sugar binding site. This is strengthened by
molecular dynamic studies of vSGLT, showing that passive water fluxes cross the protein through the sugar
transport pathway (20).
While the unitary conductance of SGLT1 for water
and urea is low relative to other channels, SGLT1 may
Physiol Rev • VOL
775
play an important physiological role in water and urea
transport in those cells with high levels of expression,
e.g., 250,000 SGLT1 molecules/enterocyte. Thus SGLT1
may be a major pathway for water and urea transport
across the intestinal brush-border membrane.
C. Coupled Water, Urea, and Glucose Transport
A surprising observation was that in oocytes expressing SGLT1, the addition of glucose immediately increased
the transport of water into the cell (133, 134, 145, 263).
The initial rate of sugar-coupled water transport was
1) directly proportional to the rate of Na⫹/glucose
cotransport, i.e., the rate varied with changes in membrane potential, sugar concentration, and temperature
(the activation energy for coupled water transport was
identical to that for Na⫹/sugar transport, 25–30 kcal/mol);
2) independent of the osmotic gradient and even occurred
against an osmotic gradient; 3) coupling was independent
of the ion used to drive sugar transport, Na⫹ or H⫹; and
4) the coupling coefficient varied by cotransporter subtype, e.g., rabbit SGLT1, human SGLT1, NIS, and a plant
H⫹/amino acid cotransporter (AAP5) where the coupling
ranged from 50 to 425 water molecules per turnover. This
implied that water coupling was related to transporter
architecture and not simply the rate of Na⫹ and solute
transport. While these experiments clearly demonstrated
that there is a close relationship between Na⫹/glucose
cotransport and the initial rate of water transport, controversy remains as to the interpretation of the results.
We have favored the water cotransport hypothesis, while
others favor a strict osmotic coupling, i.e., water flowed in
response to the osmotic gradients set up by Na⫹/sugar
cotransport into the cell (15, 109, 133, 263–265). While
there is no doubt that osmotic gradients do contribute to
coupled water flow through SGLT1, the debate centers on
the importance of unstirred layers during the initial (1–5
s) turning on of Na⫹/glucose cotransport. Potential unstirred layer effects are not likely to account for the
observed coupling between urea and Na⫹/glucose cotransport (115). In molecular dynamic studies of vSGLT, 70 – 80
water molecules accompany galactose as it moves from
the binding site into the intracellular space (20), consistent with the water pump hypothesis. Nevertheless, it is
clear that there is coupling, direct or indirect, between
water and Na⫹/glucose cotransport, and this is expected
to play a major role in water transport across the intestinal brush border (133).
D. Glucose Sensor
Until 2003, it was widely assumed that all SGLT genes
are expressed in intestinal and renal epithelial cells,
where they are responsible for the accumulation of glu-
91 • APRIL 2011 •
www.prv.org
776
WRIGHT, LOO, AND HIRAYAMA
cose and galactose. However, hSGLT3 has proven not to
be a Na⫹/glucose cotransporter, but instead a glucose
sensor (35). The gene was initially found to be expressed
in the human small intestine and muscle, and subsequent
mRNA protection assays found that it was also expressed
in uterus, lung, brain, spleen, thyroid, kidney, and trachea
(Table 1). In the small intestine, immunocytochemical
studies demonstrated that the hSGLT3 protein was present in the enteric neurons and not in intestinal epithelial
cells. In skeletal muscle, the protein was confined to the
neuromuscular junction. When expressed in Xenopus laevis oocytes, hSGLT3 was efficiently inserted into the
plasma membrane, but it was unable to transport glucose.
Electrophysiological assays revealed that glucose
caused a specific, phlorizin-sensitive, Na⫹-dependent depolarization of the membrane potential of up to 12 mV.
Radioactive glucose uptake assays under voltage clamp
revealed that the glucose-induced inward currents were not
accompanied by glucose transport. The D-glucose affinity for
hSGLT3 was low (Km 20 mM), and D-galactose was not
recognized. On the other hand, imino sugars, such as 1-deoxynojirimycin, were found to be highly potent agonists (Km
4 ␮M; Ref. 235). Based on these electrophysiological properties and the expression pattern of the gene, we have
speculated that hSGLT3 is a glucose sensor and not a Na⫹/
glucose cotransporter. This has been supported by studies
that show glucose regulates intestinal motility in rodents
and humans (103, 181) and that guinea pig enteric neurons
are reported to be glucose sensitive (121).
Further strong evidence in support of SGLT3’s function as a glucosensor came from a recent study where the
gene was expressed in C. elegans ASK chemosensory
neurons (9). On the basis of assays on chemotaxis plates,
the transgenic animals were repulsed or attracted to 10
mM glucose depending on the pH, and the responses were
blocked by 100 ␮M phlorizin. Wild-type worms were neither attracted nor repulsed by glucose.
E. Glucose Sensing in the Gut
Glucose sensors are present throughout the body,
especially in the gastrointestinal tract. These range from
the “sweet” receptors in the tongue to the glucose sensors
in pancreatic beta cells and in the small intestine. These
sensors are intimately involved in food intake, from coordinating the physiological response to an ingested carbohydrate meal, to regulating blood glucose levels, and regulating
glucose transporter expression in response to changes in
diet. It has long been recognized that one of the principal
roles of the duodenum is to act as a sensory organ (see Ref.
180). Specifically, glucose in the duodenum inhibits gastric
emptying, increases intestinal motility, and stimulates the
release of glucose-dependent insulinotropic polypeptide
(GIP), glucagon-like peptide-1 (GLP-1), and serotonin (5-HT)
Physiol Rev • VOL
from endocrine cells. The major function of GIP and GLP-1
is to stimulate pancreatic insulin secretion, whereas 5-HT
acts to modulate enteric reflexes.
The glucose sensors lining the intestine have properties similar to those of SGLTs: 1) nonmetabolized
substrates, 3-O-methyl glucoside and ␣-methyl-glucopyranoside, mimic glucose in reducing gastric emptying
and stimulate the release of 5-HT and GLP-1 from the
gut and 5-HT and GLP-1 from neuroendocrine cell models; 2) 5-HT and GLP-1 release are blocked by phlorizin;
and 3) SGLT3 is expressed in an intestinal neuroendocrine cell line (57). It is now recognized that at least
some of the intestinal glucose sensors are G proteincoupled receptors (GPCRs) found in enteroendocrine
cells (40). Perhaps the strongest evidence for this is
that mice lacking T1T3 and ␣-gustducin showed no
upregulation of SGLT1 with increasing dietary carbohydrates. The finding that glucose sensors are expressed in different cells compared with the transporter suggests a chemical signaling pathway between
cells, and this is consistent with neurogenin-3 mutations in human subjects, where a lack of enteroendocrine cells results in severe malabsorption (236).
F. Glucose Sensing in the Brain
The SGLT3 story now raises the possibility that
other SGLTs may behave as glucose sensors especially
since these genes are expressed throughout the body
(Table 1). This is of particular relevance in the brain,
where specialized neurons controlling sleep, appetite,
and hormone secretion behave as glucose detectors
(for review, see Ref. 56). There are two types of glucose-sensing neurons, glucose-excited and glucose-inhibited neurons in the hypothalamus and brain stem,
whose rate of firing either increases or decreases in
response to extracellular glucose. While the current
dogma is that these glucose responses are mediated via
KATP as in pancreatic ␤-cells, there is emerging evidence that SGLTs are involved. Namely, glucose-excited neurons are also stimulated by nonmetabolized
SGLT substrates, such as ␣-MDG, and the response is
blocked by the SGLT inhibitor phlorizin and the absence of Na⫹ (Fig. 33) (154).
There is also RT-PCR evidence that SGLT1 and -3 are
expressed in the hypothalamus (see also section V). Given
the coexpression of SGLTs and GLUT3 in neurons, it is
unlikely that the SGLTs in neurons are simply required for
glucose uptake. It is more likely that these cells respond to
changes in ambient glucose concentrations by depolarization of the membrane potential (see Fig. 3). The ␣MDGstimulated firing of glucose-excited neurons is probably mediated by an increase in Ca2⫹ influx. In summary, SGLT
genes may have important physiological roles throughout
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
FIG. 33. Response of glucose-excited hypothalamic neurons to
␣MDG, phlorizin, and Na⫹. The intracellular [Ca2⫹] of cultured hypothalamic neurons was measured based on the ratiometric images of
intracellular fura 2-AM (340/380 nm excitation, ⬎510 nm emission).
A: the responses to the addition of 12 mM glucose (G) in the presence
and absence of 100 nM phlorizin (Phl). B: the responses to glucose and
␣MDG in the presence and absence of Na⫹. C: histogram of pooled data
from A and B with the number of cells tested in the presence and
absence of phlorizin or Na⫹. [From O’Malley et al. (154).]
⫹
the body not only as Na /glucose cotransporters but as Na⫹
uniporters, water and urea channels, and glucose sensors.
XIV. PHYSIOLOGY AND PATHOPHYSIOLOGY
A. Regulation of Expression
Ongoing studies are examining the regulation of
SGLT1 and SGLT2 expression at the mRNA level, mostly
in the kidney and intestine in diabetes and as a function of
diet. The promoters of human and sheep SGLT1 have
been mapped (143, 229) and, at least in the STC-1 cell line,
sheep promoter activity is activated by glucose through a
PKA pathway (41). Both the SGLT1 and SGLT2 promoters
contain binding sites for human hepatocyte nuclear factor
1 (HNF-1) that enhance promoter activity (143, 172).
Knockout mice (HNF1␣⫺/⫺) suffered from a renal Fanconi syndrome where there was severe glucose, phosphate, and amino acid urinary wasting but there was no
intestinal phenotype (172). The glucosuria was caused by
a defect in SGLT2 expression. These results suggest that
in the mouse HNF1␣ regulates renal SGLT2 expression
but not intestinal SGLT1 expression.
Unfortunately, a lack of well-characterized antibodies means that there are a limited number of studies on
Physiol Rev • VOL
777
the regulation of protein expression in either the intestine
or kidney. Arguably one of the most careful studies is that
on the expression, SGLT1 and GLUT2 in the intestinal
biopsies of diabetic and control human subjects (43).
There was a three- to fourfold increase in SGLT1 and
GLUT2 mRNA and protein levels, and the activity of Na⫹/
glucose cotransport in brush-border membrane vesicles
also increased threefold. The molecular mechanism underlying this increase is unknown and is complicated by
the fact that mature enterocytes normally express SGLT1,
and epithelial turnover occurs with a half-time of 3– 4
days. In both the intestine and kidney of animals, the
changes in SGLT1 and SGLT2 mRNA levels with different
experimental protocols (diet, aging, diabetic model) were
generally less than fivefold.
A notable exception is the change in intestinal transport activity and protein level in sheep, where the level of
glucose transport decreased by two orders of magnitude
on weaning (201, 243). Both the rate of Na⫹/glucose
cotransport and SGLT1 protein in brush-border membranes drop by 200-fold when lambs are weaned and
dietary carbohydrate fails to reach the small intestine,
while there is a modest fourfold reduction in SGLT1
mRNA (114). Surprisingly, infusion of glucose into the
intestine of adult sheep increased the level of transporter
activity 100-fold, but the mRNA levels only increased
2-fold. This suggested translational or posttranslational
regulation of the transporter in brush-border membrane.
Infusion of 30 mM glucose, fructose, or 2-deoxy-glucose
into the intestine of the ruminant sheep for 3 h returned
the level of brush-border Na⫹/glucose cotransport and
protein to the preruminant state after several days (42).
This body of work has led to the conclusion that SGLT1
upregulation is mediated by a glucose sensor that initiates
a signaling pathway linked to a cAMP-PKA pathway that
takes several days to act.
Over the shorter term of minutes, activation of the
cAMP pathway results in the recruitment of an intracellular pool of transporters to the plasma membrane. The
experiments were performed on rabbit SGLT1 expressed in Xenopus laevis oocytes. With the use of
membrane-permeable 8-BrcAMP to activate PKA, there
was a reversible 30% increase in the maximum rate of
Na⫹/glucose transport that occurred within minutes,
whereas activating PKC with membrane-permeable sn1,2-dioctytanoylglycerol (DOG) resulted in a decrease
in the maximum rate of transport by 60% (72). There
were no changes in the kinetics of SGLT1 apart from
the increase in the maximum rate of transport. These
changes in maximum transport rate were due to
changes in the number of transporters in the plasma
membrane. Both 8-BrcAMP and DOG increased the
number of hSGLT1 molecules in the membrane, as did
calyculin A, an inhibitor of phosphatases 1 and 2A.
Even though there are four PKC and one PKA consen-
91 • APRIL 2011 •
www.prv.org
778
WRIGHT, LOO, AND HIRAYAMA
sus sites in human SGLT, we have suggested that this
regulation occurs by regulated endo- and exocytosis
and not phosphorylation of SGLT1.
Figure 34 shows thin-section and freeze-fracture
electron micrographs of the 120-nm vesicles in the cytoplasm fusing with the plasma membrane of Xenopus laevis oocytes (248). Note the increase in intramembrane
particles (SGLT1 proteins) in the SGLT1 expressing
oocytes. The freeze-fracture images of vesicles isolated
from the cytoplasm of SGLT1-expressing oocytes show
that there are 10 –20 SGLT1 proteins per vesicle. Given the
rate of cotransporter insertion into the membrane
(250,000 s⫺1), we estimate that 10,000 of these vesicles
fuse with the oocyte plasma membrane per second. Since
the membrane area increased by only 60 ␮m2/s, equivalent to fusion of 1,500 120-nm vesicles, we suggest that
concurrent endocytosis accounts for the smaller increase
in area. With an average number of transporters in the
plasma membrane before stimulation of 2. 5 ⫻ 1011/
oocyte, a 30% increase 30 min after 8-BrcAMP treatment
represents an insertion rate of 1 ⫻ 107 SGLT1/s. For the
area of an oocyte plasma membrane, 30 ⫻ 106 ␮m2, this is
equivalent to a fusion rate of 1 vesicle·␮m⫺2·s⫺1 (248).
This rapid insertion of SGLT1 into the plasma membrane
indicates an intracellular reserve pool of vesicles within
the cell poised to fuse with the cell membrane. There is
evidence that cAMP and theophylline increase the maximum velocity of SGLT1 activity in enterocytes within 20
min (197), presumably by a similar mechanism.
One factor that may be involved in the regulation of
transcription and trafficking of SGLT1 in the intestine is
the regulatory gene RS1 (98, 101, 232, 233). In RS10 null
mice, SGLT1 was upregulated and the mice became
obese, apparently due to an increase in insertion of
SGLT1 vesicles from the trans-Golgi network.
B. Intestinal Absorption
The currently accepted dogma is that glucose and
galactose absorption occurs across mature enterocytes in
two stages: the first is the uphill accumulation of these
hexoses across the brush-border membrane by SGLT1,
and the second is the downhill transport from the cell into
FIG. 34. Vesicle trafficking in oocytes.
Thin-section electron micrographs (A and
B) and freeze-fracture electron micrographs (C and D) showing membrane vesicles in Xenopus laevis oocytes expressing
rabbit SGLT1. A and B show 120-nm-diameter vesicles approaching and fusing with
the plasma membrane, and C and D show
the P-face (P) of similar vesicles containing
7.5-nm-diameter SGLT1 intramembrane
particles (see also Fig. 2). Cyt, cytoplasm.
Magnification, ⫻150,000; scale bar, 0.15
␮m. (248). E shows a freeze fracture of
membrane vesicles isolated from oocytes
expressing rabbit R427A-SGLT1 (a trafficking mutant, Ref. 135) using density gradient
centrifugation. The fraction containing
R427A-SGLT1 was identified using Western
blotting. (From B. A. Hirayama, M. Kreman, G. Zampighi, and E. M. Wright, unpublished data.)
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
779
FIG. 35. Model for glucose and galactose transport across enterocytes. Glucose (and galactose) is accumulated within the cell from the
gut lumen by SGLT1 in the brush-border membrane. The energy for
uphill sugar transport is provided by the sodium electrochemical potential gradient across the brush-border membrane. The two sodium ions
that enter the cell with each sugar molecule are pumped out across the
basolateral membrane by the Na⫹-K⫹ pump. Sugar that accumulates
within the cell exits across the basolateral membrane into the blood.
glucose absorption equally well in wild-type and GLUT2null mice. In addition, in vivo experiments in rats showed
that ␣MDG was absorbed across the intestine with kinetics similar to D-glucose and D-galactose (31) despite the
fact that ␣MDG is not a substrate for GLUT2 (252). This
suggests that a mechanism other than facilitated diffusion
through GLUT2 is responsible for exit from the enterocyte.
One yet-to-be explained result was the finding that absorption of 3-O-methyl-D-glucoside was abnormal in the
GLUT2⫺/⫺ mice (206).
These observations on GLUT2⫺/⫺ mice and FBS
patients raise doubts about the popular hypothesis that
GLUT2 plays an important role in the regulation of
glucose absorption across the intestine (see Ref. 89). A
central tenet of this hypothesis is that in response to a
glucose meal, GLUT2 is redirected into the brush-border membrane to account for the increase in sugar
absorption. This is inconsistent with normal glucose
absorption and regulation by cAMP in GLUT2-null mice
and normal glucose absorption in FBS patients. There
was a fundamental flaw in the design of Kellett’s exper-
blood through GLUT2 present in the basolateral membrane (Fig. 35). Soon after the cloning of SGLT1 from the
rabbit small intestine, it was shown using both Western
blotting and immunocytochemistry that the protein is
only expressed in the brush-border membrane (70, 209).
The NH2-terminal domain of SGLT1 (amino acids 1–19)
appears to detemine the targeting of the protein to the
brush-border membrane (207). There was some supranuclear positive staining in the enterocytes on the
mid-villus region consistent with the biosynthetic pathway. Immunostaining of the human brush border is
shown in Figure 36A.
Perhaps the most definitive evidence for the role of
SGLT1 in sugar transport across the human brush-border
membrane came from studies of the disorder glucose
galactose malabsorption (see below), the result of mutations in the SGLT1 gene. Figure 36B shows immunocytochemical location of SGLT1 in the patient with a C292Y
mutation on both alleles. In the duodenal biopsy, the
C292Y-SGLT1 protein was trapped between the nucleus
and the brush-border membrane (Fig. 36B). Studies of the
mutant protein expressed in Xenopus laevis oocytes also
showed that there is no Na⫹/glucose cotransport and that
a trafficking defect prevents insertion of the mutant protein into the plasma membrane (142).
Although GLUT2 is expressed in the basolateral
membrane of enterocytes, the role of this protein in glucose absorption is unclear based on studies of GLUT2⫺/⫺
mice and patients with the Fanconi Bickel Syndrome
(FBS), caused by mutations in GLUT2 (189, 206). Oral
glucose tolerance tests were normal in both GLUT2⫺/⫺
knockout mice and FBS patients and cAMP increased
FIG. 36. SGLT1 immunostaining of intestinal biopsies from a control subject (A) and a glucose galactose malabsorption (GGM) patient
with the homozygous C292Y-SGLT1 mutation (B). In each section, the
nuclei are counterstained with DAPI. (From P. Lostao, B. A. Hirayama,
and E. M. Wright, unpublished data.)
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
780
WRIGHT, LOO, AND HIRAYAMA
iments, where they dissected the contributions of
SGLT1 and GLUT2 to glucose absorption in rats on the
basis of the phloretin inhibition (90); they assumed that
1 mM phloretin only inhibited GLUT2; and the phloretin-insensitive fraction of glucose absorption was
through SGLT1. It has been well known that phloretin is
a potent noncompetitive SGLT1 inhibitor with a Ki of 50
␮M, e.g., Refs. 66, 159. This means that Kelletts’s laboratory grossly underestimated the importance of
SGLT1, and this is reinforced by the observation that
glucose absorption can be totally defective in patients
with mutations in SGLT1.
C. Oral Rehydration Therapy
Perhaps the most important medical application of
SGLT1 is its critical role in the treatment of secretory
diarrhea in children, the elderly, and immunocompromised patients. According to the World Health Organization (WHO), diarrhea caused by cholera is the second
leading cause of childhood mortality, killing some 1.5
million children ⬍5 years of age. Cholera infection
causes massive diarrhea, resulting in dehydration that
can lead to death unless the patient is rehydrated,
which in adults may require 80 liters of intravenous
fluids over 5 days (73). The challenge of such heroic
fluid replacement to save the patient was overcome by
the discovery that one could simply administer an oral
rehydration solution (ORS) containing salt and glucose.
The glucose in the ORS is absorbed along with two
sodium ions by SGLT1 on the brush-border membrane,
and water is either cotransported with the sugar and
sodium, or follows by osmosis (see sect. XIIIC). We
estimate that, directly or indirectly, SGLT1 can promote
about 6 liters of water absorption daily in the normal
adult intestine. The current WHO/UNICEF oral rehydration solution contains 75 mM NaCl, 75 mM glucose, 20
mM KCl, and 10 mM sodium citrate with an osmolarity
of 245 mosM, and commercial formulations in the
United States and Europe are usually flavored to make
the solution more palatable. In the field where ORS is
not available, there are several homemade recipes that
use locally available supplies of carbohydrate, see,
http://rehydrate.org. In all cases, the ORS is designed to
increase intestinal salt, sugar, and water absorption to
match or exceed the secretory diarrhea caused by the
toxin (see Fig. 35).
Since the first successful clinical trials of ORT in
1968 at the Cholera Research Laboratory in Dhaka (74,
171), the number of deaths has decreased dramatically,
and this therapy has been extended to even include
diarrhea caused by other pathogens, including rotavirus, E. coli, and Yersinia infections (179) in both developed and underdeveloped counties worldwide. In
Physiol Rev • VOL
1978, an editorial in the British medical journal The
Lancet (2: 300 –301) hailed ORT as the “most important
medical advance of this (20th) century.” Readers interested in the historical steps and missteps in the development of oral rehydration therapy are referred to the
review by Ruxin (186).
D. Glucose Galactose Malabsorption
GGM was first described in two reports in 1962 as a
watery and severe diarrhea in newborn children that is
fatal within weeks unless lactose, glucose, and galactose
are removed from the infant’s diet (108, 119). The diarrhea
ceases immediately on removing sugar and promptly resumes when lactose, glucose, or galactose are added back
into the diet (250). It was predicted that GGM is due to a
defect in the brush-border Na⫹/glucose cotransporter.
This was confirmed by Eric Turk following the cloning of
human SGLT1 (64, 222), when a homozygous mutation,
D28N, was found in two sisters with GGM, and that each
parent was a carrier for this mutation. In Xenopus laevis
oocytes, the D28N SGLT1 protein was shown to be unable
to transport glucose. Prenatal screening in two subsequent pregnancies in this large consanguineous family
revealed that the probands’ sibling was a carrier and a
cousin did not carry the mutation (141). This study also
confirmed that GGM was an autosomal recessive disorder. Patients with GGM remain intolerant of glucose and
galactose for the rest of their lives, and our oldest subject,
now 58, lives an apparently normal healthy life on a
sugar-free diet. His blood chemistries are within normal
limits, but malabsorption symptoms return immediately
on ingesting food with the offending sugars. There have
been a number of reports of nephrocalcinosis in GGM
children (see Ref. 203). It is thought that this may arise
due to hypercalcemia, metabolic acidosis, and dehydration as renal tubule dysfunction, and these symptoms
often resolve on removing glucose and galactose from the
diet.
Over 80 patients with a diagnosis of GGM have been
screened for mutations in the SGLT1 gene (55, 87, 105,
140, 203, 213). The most clear diagnosis of GGM includes
observations that 1) the diarrhea returns when glucose or
galactose, but not fructose, is added to the diet; 2) oral
glucose tolerance tests produce flat blood glucose curves;
and 3) positive glucose hydrogen breath tests indicate
glucose malabsorption. In all but a few patients, mutations have been indentified in SGLT1 that cause the defect
in sugar absorption (87, 142, 250). In several cases where
no mutations were found, the diarrhea was found to be
due to the novel disorder enteric anendocrinosis due to
mutations in the neurogenin 3 gene (236). This indicates
that signals from endocrine cells are required for normal
absorption.
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
About 65% of GGM patients tested have the same
mutation on each allele, and the remainder have different mutations on each allele, i.e., they are compound
heterozygotes. There are only a handful of cases where
the same mutation is found in unrelated families, but
these families are of similar ethnic backgrounds. This
may reflect an increase in the frequency of these variants in a given population (see below). The majority of
mutations are missense, but nonsense, frame shift,
splice-site, and promoter mutations have also been documented. The nonsense, frame shift, and splice site
mutations all produce truncated proteins that are predicted to be unstable and/or nonfunctional. To determine whether a mutation causes an actual defect in
glucose and galactose transport or is simply a polymorphism, it is essential to measure the transport activity
of the mutant protein. This has usually been carried out
using the Xenopus laevis expression system (55, 87,
140, 142, 222). We have confirmed that two truncated
proteins, Y191X and R359X, are nonfunctional when
expressed in Xenopus laevis oocytes (142). The missense mutations are distributed throughout the protein
(see Fig. 37) with no particular “hot spots.” Of the 27
mutations tested, all but 3 show dramatically reduced
Na⫹/glucose transport activity, and the most common
cause is a failure in the insertion of the protein into the
plasma membrane as determined from charge (Qmax)
measurements, freeze-fracture electron microscopy,
781
and/or immunohistochemistry. In several patients with
these homozygous mutations, we have also found using
intestinal biopsies that the transporter is not inserted
into the brush-border membrane of enterocytes (see
Fig. 36). There are three nonfunctional mutant proteins
that are inserted into the plasma membrane, R135W,
Q457R and T460P, and one of these, Q457, has been
found to be important in binding sugar (38) (see Fig.
17). The residues that are mutated to produce nonfunctional proteins are conserved in the SGLT family of
proteins (see Fig. 190 –7 in Ref. 250).
Three missense mutations that did not significantly
alter the function of SGLT1, N51S, A411T, and H615Q, are
not conserved in the SGLT family of proteins, but they
were found at a high frequency, 9%, in the EuropeanAmerican population. As part of the University of California San Francisco Pharmacogenetics of Membrane Transporters Project (www.pharmacognetics.ucsf.edu), we
have examined 552 alleles in the Coriell Institute genomic
DNA collection for variations in hSGLT1 (Fig. 38). Only 12
nonsynonymous mutations were found, and only 6 are at
a frequency of ⬎1%. Although these three common variants were found in our pool of GGM patients, it was
second mutations that produced the defect in glucose
transport. None of the GGM mutations has been found in
this sample of 552 alleles, as expected from the rarity of
the disorder.
FIG. 37. Missense mutations of SGLT1 identified in patients with GGM. The mutations are shown on a secondary structure model with the
transmembrane helices from the vSGLT structure highlighted in gray. [Revised from Wright (244).]
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
782
WRIGHT, LOO, AND HIRAYAMA
FIG. 38. Intronic changes in the hSGLT gene in 552 alleles. The 22 variants include 12 nonsynonymous changes. The frequency of each
polymorphism in each population is included in parentheses. CA, Caucasian; AA, African American; AS, Asian; PA, Pacific Island. Synonymous
variants are shown in green.
E. Enteric Infection
F. Renal Reabsorption
There is emerging evidence that upregulation of intestinal SGLT1 may afford protection against enteric infections.
High glucose concentrations (25 mM) protected SGLT1
transfected Caco-2 cells against apoptosis induced by bacterial lipopolysaccharide (LPS) (259). Furthermore, LPS increased glucose transport activity in these cells. These studies have been applied to Giardia duodenalis-induced apoptosis in this system, and extended to demonstrate that G.
duodenalis proteolytic fragments increased SGLT1 expression in the apical membrane (260). Arguably, the most convincing demonstration of the cytoprotective effects of
SGLT1 was the demonstration that oral glucose gave 100%
protection against lethal endotoxic shock in mice (160).
Lethal endotoxic shock was induced by LPS and pretreatment with oral glucose, or the nonmetabolized SGLT1 substrate 3-O-methyl-D-glucose, resulted in 100% survival of
mice against the insult. These studies suggest a novel immunological role for SGLT1 and new approaches to managing
severe sepsis infections that result in chronic diseases such
as inflammatory bowel disease (IBD).
Glucose is freely filtered across the renal glomerulus,
amounting to 180 g·day⫺1·1.73 m⫺2 in adults (for a review,
see Ref. 245). In healthy individuals, ⬍0.5 g/day is lost in
the urine, so ⬎97% of the filtered load is reabsorbed. The
urine remains virtually free of glucose within the normal
range of blood (plasma) glucose levels (4 –10 mM), but as
the plasma concentration exceeds ⬃14 mM, glucose appears in the urine and above 20 mM the excretion increases in proportion to the filtered load. In other words,
the reabsorption of glucose from the glomerular filtrate
saturates at ⬃425 g/day with an apparent Km of ⬃10 mM.
The most common cause of glucose in the urine of patients is due to the high blood glucose levels associated
with diabetes. Micropuncture studies of frog and rat
nephrons have demonstrated that the glomerular filtrate
is glucose-free when it reached the end of the proximal
tubule (Fig. 39, A and B). Microperfusion experiments
with isolated nephrons from rabbits established that active glucose reabsorption occurs in the “early” convoluted
proximal tubule via a low-affinity (Km 2 mM) high-capac-
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
783
FIG. 39. Glucose reabsorption from the proximal tubule. A: anatomical arrangement of a kidney nephron with its blood supply. B: summary of
the results of micropuncture experiments to measure the glucose concentration as fluid flows freely along the tubule from the glomerulus (244).
C: a model for glucose reabsorption across the epithelium of the proximal tubule. Similar to that in enterocytes (Fig. 34), the Na⫹ electrochemical
potential gradient provides the energy source, and transport across the epithelium occurs in two steps. Glucose is transported across the apical
membrane by SGLT1 and SGLT2. SGLT2 is predominantly located in the S1 and S2 segments and SGLT1 is in S3. GLUT2 is the major glucose
transporter across the basolateral membrane.
ity system (80 pmol·min⫺1·mm⫺1) system, whereas active
reabsorption in the late proximal tubule occurred by a
high-affinity (Km 0.5 mM), low-capacity system (10
pmol·min⫺1·mm⫺1) system.
Further evidence for two transport systems in the
human kidney came from brush-border membrane vesicle
experiments where there was a hint of low- and highaffinity Na⫹/glucose cotransport systems. Additional studies on brush-border membranes from the outer cortex and
outer medulla of rabbit kidney provided additional support for two different SGLTs: one in the outer cortex had
a glucose Km of 6 mM and one in the outer medulla with
a Km of 0.3 mM, and the low-affinity one was more sensitive to phlorizin than the high-affinity one. These two
transporters have come to be called SGLT2 and SGLT1
(245) (Fig. 39), and their properties are in close agreement with those of the cloned SGLT1 and SGLT2 proteins
(Table 1) (81).
The cellular mechanism of renal glucose reabsorption is shown in Figure 39C. The two-step process is
similar to that in enterocytes with uphill transport across
the brush border into the epithelium and downhill facilitated diffusion from the cell into the blood across the
basolateral membrane. Antibodies against the cloned rabbit SGLT1 were used to examine the location of SGLT1 in
the rat kidney (22, 210), and it was found that the protein
was present in the brush-border membrane of all three
segments of the proximal tubule. This has been confirmed
recently with another antibody (4). In proximal tubule,
GLUT1 was only found in the basolateral membrane of
the S3 segment (210). Basolateral GLUT2 colocalizes in
cortical tubules expressing brush-border SGLT1 (22). Immunohistochemistry localized SGLT2 to the brush border
of the early proximal tubule in mice, and this was absent
in SGLT2⫺/⫺ animals (227). The SGLT2⫺/⫺ mice exhibPhysiol Rev • VOL
ited glucosuria, and free-flow micropuncture studies
showed that there was no glucose reabsorption in the
early proximal tubule.
Unlike the small intestine, GLUT2 is very important
in the normal reabsorption of glucose in the kidney, as
patients with FBS exhibit massive renal glucosuria (244).
Those patients who have a homozygous premature stop
codon in the GLUT2 gene had a renal glycosuria of up to
200 g/day. This indicates that GLUT2 is the major basolateral glucose transporter involved in the reabsorption of
the glucose filtered load. Basolateral GLUT1 and GLUT2
are also involved in the entry and exit of glucose into
other segments of the nephron, e.g., in the release of
glucose produced by gluconeogenesis.
Mutations in brush-border SGLT1 and SGLT2 also
cause renal glucosuria, but in the case of SGLT1, the
glucosuria that accompanies GGM is mild, e.g., one patient with severe SGLT1 mutations (G426R and S159P)
lost only 1.5 g glucose/day in the urine (244).
G. Familial Renal Glucosuria
This is a rare autosomal recessive disorder where
glucose is excreted into the urine, ⬎1 g/day, when blood
glucose levels and oral glucose tolerance tests are normal
(244). There are no other renal abnormalities, even when
the glucose excretion exceeds 100 g/day, and there are no
other systemic consequences. In the case of one patient
with severe renal glucosuria (⬎109 g/day), who was reevaluated 20 years after glucosuria was first detected, there
were no chronic nephrological complications (194). Table 6
contains a summary of the glucosuria and the SGLT2 mutations in five patients with severe FRG (13, 190, 244).
Glucose excretion ranges from 1 to 162 g·1.73
m⫺2·day⫺1, and the mutations are those expected for an
91 • APRIL 2011 •
www.prv.org
784
TABLE
WRIGHT, LOO, AND HIRAYAMA
6. Familial renal glucosuria
Subject
Glucose Excretion,
g·1.73 m⫺2·day⫺1
Allele 1
Allele 2
01-1
06-1
15-1
15-2
17
⬎126
69
202
80
30-92
P347X
W440X
R137H
R137H
Y128X
P347X
W440X
⌬385-8
⌬ 385-8
Y128X
Renal glucose excretion and SGLT2 mutations in patients with familial
renal glucosuria (13, 190, 244).
autosomal recessive inheritance: 12 homozygous and 7
compound heterozygous that include premature stops,
frame shifts, and missense mutations. So far, none of the
FRG mutations has been tested for functional SGLT2
effects, due to the low expression of SGLT2 in heterologous expression systems, so it is difficult to relate the
mutations to the severity of the glucosuria. However, it is
noteworthy that 1) in subjects with severely truncated
proteins, one excretes most of the filtered load (374X)
while two others reabsorb about half of the filtered load
(W440X and Y128X) (Table 6); and 2) there is a huge
discrepancy in the glucosuria between two siblings with
the same mutations. In subject 15–1, the glucose excretion is 202 g·1.73 m⫺2·day⫺1, while in subject 15–2 it is
only 80 g·1.73 m⫺2·day⫺1. This, combined with the 50%
inhibition of glucose reabsorption by SGLT2 inhibitors
(95), suggests that SGLT2 may not be solely responsible
for glucose reabsorption in the kidney. Phlorizin is
known to inhibit reabsorption completely (16), and this
points to the involvement of another SGLT. It is unlikely this is SGLT1, as the data suggest this transporter
accounts for ⬍10% of the normal amount absorbed.
Another possibility is SGLT5, expressed exclusively in
the renal cortex (Table 1).
SGLT2 mutations found in subjects with FRG are
shown on a secondary structure model in Figure 40.
These are distributed throughout the protein in both
transmembrane domains and hydrophilic loops, and none
is at the predicted sugar or Na⫹ coordination sites, although F453L is at a predicted outer gate residue (see
above). Several SGLT2 mutations occur in similar locations to GGM mutations in SGLT1, e.g., R137H, R300C,
G304K, and R499H (Fig. 37). While the effect of mutations
on SGLT2 function has not yet been reported, we note
that in SGLT1 mutations at R135W, R300S, A304V, and
R499H produce defects in glucose transport, largely
through trafficking problems (142, 250). The truncated
SGLT2 proteins, 128X, 186X, 347X, and 440X, are not
expected to produce stable functional proteins.
FIG. 40. The location of SGLT2 mutations on a secondary structure model of the protein. The gray circles indicate the location of TM2-TM14
based on the crystal structure of vSGLT (48). Missense mutations are highlighted in yellow, premature stops in blue, and deletions in red. As with
GGM mutations in SGLT1, the mutations are distributed throughout the protein. [Modified from Wright (244).]
Physiol Rev • VOL
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
In summary, SGLTs play a major role in salvaging glucose from the glomerular filtrate to avoid loss of this nutrient to the urine. However, in terms of renal function, this is
not deemed essential as subjects with the rare disorder
known as FRG do not have any chronic kidney complications, nor do they become hypoglycemic. This has encouraged the pharmaceutical industry to develop SGLT2 inhibitors to lower blood glucose levels in diabetics (see below).
H. Bile, Milk, and Saliva
It has long been known that glucose enters the bile
in the liver and then is reabsorbed in the interhepatic
bile ducts by SGLT1 in the apical membrane and GLUT1
in the basolateral membrane of cholangiocytes (111).
This was confirmed in studies of glucose transport in
microperfused rat intrahepatic bile ducts and further
demonstrated that glucose transport was accompanied
by water transport (144). In vivo it was reported that as
the rate of glucose absorption increased there was a
reduction in bile flow, and when glucose absorption
was inhibited by phlorizin, the bile flow increased. This
relationship between glucose and bile flow may contribute to the low bile flow, and associated symptoms
such as jaundice, in diabetes.
Glucose supply to the lactating mammary gland is
important in providing energy and as precursor for lactose production and secretion. It has been argued that
glucose transporters in the basolateral membrane of
mammary epithelia are essential for milk production. The
expression of glucose transporters in the human lactating
mammary gland was examined in a novel manner by
Obermeier et al. (155) when they examined the genes
expressed in epithelial cells isolated from fresh milk.
They found SGLT1 mRNA by RT-PCR and SGLT1 protein
by Western Blot analysis, but did not find GLUT1. SGLT1
mRNA and protein was also found in rat lactating mammary gland tissue (198). A more complete analysis of
glucose transporter gene expression has been reported
for the bovine mammary gland (266). The largest increase
in expression, from ⫺40 to ⫹7 days after parturition, was
for GLUT1 and GLUT8 (10- to 100-fold), while there were
more modest increases in SGLT1 and SGLT2 expression.
At this time, we are unaware of any studies on the distribution of the SGLTs and GLUTs between the apical and
basolateral membranes. Likewise, we are unaware of any
mammary gland phenotype in women with severe mutations in SGLT1 (GGM) or SGLT2 (FRG).
There has been an increasing awareness of oral
health complications in diabetes and the possible involvement of the salivary glands (187). Impaired salivary gland function has been reported in diabetes, including low rates of flow and high glucose concentrations. SGLT1 was first reported to be expressed in the
Physiol Rev • VOL
785
basolateral membrane of sheep parotid glands (212)
and more recently in the apical membrane of intercalated ducts of the rat submandibular gland (4). We have
also found SGLT1 in the apical membrane of human
sublingual glands (Hirayama and Wright, unpublished
data). What is the functional significance of SGLT1
expression in salivary glands, glucose secretion by the
acinar cells, and glucose reabsorption in the ducts? It is
possible that glucose reabsorption from the duct occurs by a two-stage process, i.e., apical Na⫹/glucose
cotransport through SGLT1 followed by downhill transport across the basolateral membrane through a GLUT;
this should be tested in microperfusion studies on isolated ducts from control and diabetic animals. Again, it
would be informative to evaluate salivary secretion in
subjects with GGM and with FRG.
I. Cancer
Glucose is a major source of energy, and the demand
for glucose in cancer cells is even higher than normal
cells. This is the basis for the detection and staging of
tumors using 2FDG (see sect. VC). However, some tumors
do not accumulate 2FDG, a substrate for GLUTs but not
SGLTs, increasing interest in the expression of SGLTs in
cancer. Inspection of the EST databases (www.ncbi.nlm.
nih.gov/unigene) indicates that SGLT1 is expressed in
colorectal, head and neck, and prostate tumors, and
SGLT2 is expressed in colorectal, gastrointestinal, head
and neck, and kidney tumors as well as in chondrosarcomas and leukemia. There are a handful of publications on the mRNA levels of SGLT1 and SGLT2 and
immunohistochemistry of SGLT1 in primary tumors,
and metastatic lesions of lung, pancreatic adenocarcinomas, and head and neck cancers (65, 84). SGLT1 was
expressed in well-differentiated squamous cultures of
head and neck carcinomas, SGLT2 was expressed in
metastatic lesions of lung cancers, and SGLT1 protein
was reported to be expressed in primary pancreatic
adenocarcinomas (14). We were unable to detect specific SGLT1 expression in several samples of other
tumors (lung, liver, testis, sarcomas, breast, colon,
prostate, and squamous cell carcinomas of the mouth).
No antibodies are yet available to screen for other
SGLTs in these and other tumors.
Recently, an intriguing role for SGLT1 in the survival
of cancer cells was postulated where epidermal growth
factor receptor (EGFR) stabilizes SGLT1 to prevent autophagic cell death (239). This may explain the resistance
of tumor cells to chemotherapeutic agents and tyrosine
kinase inhibitors. A novel strategy has also been proposed
to deliver chemotherapeutic agents into tumor cells
through SGLTs. This involves covalently linking a nitrogen mustard reagent to D-glucose though the ␤-C1-OH
91 • APRIL 2011 •
www.prv.org
786
WRIGHT, LOO, AND HIRAYAMA
group, e.g., ␤-D-glucosylisphosphoramide mustard (glufosfamide). These drugs are currently in phase I and II
trials for pancreatic adenocarcinomas and other solid
tumors (19, 200).
J. Diabetes
This chronic disease is a disorder of glucose homeostasis where blood glucose levels greatly exceed the normal levels, ⬎⬎10 mM, and there is reduction in insulin
secretion and/or sensitivity. If the hyperglycemia is left
untreated, it results in glucose toxicity, which damages
blood vessels and peripheral nerves, leading to blindness,
kidney failure, peripheral neuropathy, cardiovascular disease, and other serious complications. It is estimated that
(25 million) patients have diabetes, close to 10% of the
adult population in the United States, and the number is
growing at an alarming rate. One of the earliest symptoms
is a loss of glucose to the urine due to hyperglycemia
overwhelming the reabsorption capacity of SGLTs in the
proximal tubule (see sect. IXD). Current therapies to
combat this disease are centered on controlling blood
glucose levels by increasing insulin secretion, improving
insulin sensitivity, and reducing liver glucose output and
intestinal glucose absorption. As the disease progresses,
patients require combinations of medicines and, unfortunately, adverse side effects compromise compliance and
the health of the patient.
There is a growing interest in alternative therapies to
manage diabetic patients, and one is to control blood
glucose by inhibiting SGLT2 in the kidney. The pharmaceutical industry has been encouraged by 1) studies demonstrating that intravenous phlorizin decreased blood glucose levels without producing hypoglycemia in diabetic
animals, and this was accompanied by an improvement in
insulin resistance (185); 2) reports that FRG is a benign
disorder with no long-term renal abnormalities (see sect.
IXE); and 3) cloning of the “renal” SGLTs enabling in
vitro studies (see above). The strategy has been to develop oral SGLT2 inhibitors. Oral phlorizin does not fit the
bill as it blocks gastrointestinal absorption of glucose and
thus produces osmotic diarrhea. Furthermore, phlorizin is
hydrolyzed to glucose and phloretin by the intestinal
brush-border lactase-hydrolase, so little of the intact molecule is absorbed.
The proof of concept for SGLT2 targeted therapy was
provided by Oku et al. (156) who demonstrated that a
prodrug, T-1095, was absorbed from the gut into the circulation; this resulted in renal glucose excretion in diabetic animals and lowered blood glucose levels. T1095
also suppressed postprandial hyperglycemia and reduced
hyperinsulinemia and hypertriglyceridemia in diabetic rodents. In the decade since, there have been at least 21
SGLT2 inhibitors that entered the drug pipeline. Most
Physiol Rev • VOL
have exploited the same chemical space as phlorizin (see
sect. VIIIC), and many of these compounds are in phase
I to III clinical trials. To cite one example, we focus on the
C-aryl glucoside dapagliflozin (Fig. 19). This drug had a
SGLT2 inhibitor constant (EC50) of 1 nM and a selectivity
of ⬃1,200 for SGLT2 over SGLT1 (237).
Safety studies of dapagliflozin in healthy human subjects (95) indicated that the drug is well tolerated with
minimal adverse events and no clinically relevant changes
in vital signs relative to the placebo controls. The drug is
rapidly absorbed with plasma levels reaching a maximum
2 h after oral administration and thereafter falling with a
half-time of 17–18 h; unlike the O-glucosides, the C-arylglucosides are resistant to glucosidases. Ninety-seven percent of the drug stays bound to plasma proteins, and this
probably accounts for the low renal clearance and urinary
excretion. There was a dose-dependent excretion of glucose in urine up to a maximum of 60 – 80 grams/24 h, i.e.,
50% inhibition of renal glucose reabsorption, and this
persists with daily oral dosing for 2 wk. In these healthy
volunteers, there was no reduction in serum glucose or
insulin. The maximum reduction in glucose reabsorption
of ⬃50% is surprising since dapagliflozin inhibits SGLT2
completely in vitro (see sect. VIIIC). This is probably
related to the fact that in vivo 97% of the drug is bound to
plasma proteins and ⬍2% of the injected dose is recovered in the urine.
Phase II clinical trials with type 2 diabetic patients
for 2–12 wk and phase III trials for up to 24 wk have been
reported (51, 96, 120). In general, the highest doses of
dapagliflozin produce a sustained 30 – 65 g/day urinary
glucose excretion, a 22 ⫾ 10% reduction in fasting serum
glucose, and a 20 ⫾ 10% reduction in postprandial glucose
absorption (area under the plasma glucose concentration
curve). The loss of urinary glucose, 200 –300 kcal/day,
results in weight loss (up to 2.5 kg), increases in urine
volume (up to 470 ml/day) and hematocrit (up to 3%), and
an associated modest reduction in diastolic blood pressure of 2–5 mmHg. The results so far suggest that antiSGLT2 inhibitors may be useful in reaching the goals for
low glycemic control in type 2 diabetic patients and reducing glycosylated hemoglobin (Hb A1C) levels to ⬍7%.
According to these reports, the dapagliflozin treatment for
up to 24 wk produces no remarkable clinical side effects
relative to the placebo controls, as expected from the
long-term follow up with one patient with massive FRG
(194).
In summary, the pharmaceutical industry has in a
remarkably short time advanced the proof of concept of
SGLT2 inhibitors for managing hyperglycemia in diabetic patients to phase III clinical trials. Only time will
tell if these novel drugs will be effective and safe in
patients.
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
XV. OUTLOOK
To gain perspective into the future, we look at the
past and ask: what have we learned since 1987?
First, SGLT1 was the initial member of a new gene
family of 12 members, SLC5, only 4 of which are glucose
transporters. Second, it was discovered that GGM and
FRG, rare autosomal defects in glucose transport, are the
result of mutations in SGLT1 and SGLT2. Third, in addition to the intestine and kidney, SGLT1 and -2 are widely
distributed in the body, including brain, salivary glands,
lactating mammary gland, and bile ducts. Fourth, SGLTs
are multifunctional proteins behaving as Na⫹/glucose
cotransporters, water and urea channels, glucose sensors,
and coupled water and urea transporters. Fifth, an eightstate ordered kinetic model accounts quantitatively for
the observed steady-state kinetics of SGLT1. Sixth, SGLT
proteins can be overexpressed, purified, and reconstituted in liposomes for biochemical and functional studies.
Seventh, SGLTs belong to a structural family of transporters in different gene families with common transport
mechanisms. Eighth, the x-ray structure of vSGLT, and
related Na⫹ cotransporters, has begun to allow us to
explain Na⫹/sugar cotransport at an atomic level. Ninth,
the SGLTs are now drug targets for the treatment of
chronic diseases, e.g., SGLT2 inhibitors to control blood
glucose in diabetes.
XVI. UNRESOLVED PROBLEMS
Starting with our need to integrate structure and
function, we must obtain high-resolution structures of
human SGLTs to 1) identify the second Na⫹ binding site
in those transporters where the Na⫹/sugar coupling coefficient is 2. This together with being able to parse out the
kinetics of Na⫹ binding to each site will enable one to
understand both the fast transporter kinetics, a major
limitation with existing models, and how Na⫹ binding
alters transporter structure. 2) We must identify the origin
of the “gating charge” of these transporters and unlock
the origin of the voltage dependence of sugar transport.
3) We must determine the structure in different conformation states, e.g., Na⫹ free, Na⫹ bound, and inhibitor
bound. This will provide more frames in the movie to
understand the whole catalytic cycle. 4) We must resolve
the structure of SGLTs with high-affinity inhibitors bound.
This may explain the selectivity differences between
SGLT1 and SGLT2 inhibitors. 5) We must identify differences in structure between hSGLT1 and hSGLT3 to let us
know why hSGLT3 is a glucose sensor and not a glucose
transporter. In addition, this may identify the gate that
prevents sugar exit to the cytoplasm in hSGLT3 and the
link coupling sugar and Na⫹ in SGLT1. 6) High-resolution
structures will indentify the pathway for water and urea
through the transporters.
Physiol Rev • VOL
787
The second challenge is to understand substrate
specificity, kinetics, and function of the poorly understood SGLT4 and SGLT5 given the hints we already have
about their unique sugar specificity and/or expression
profile. There is also a need to clarify the regulation of
expression and function of alternatively spliced variants.
A third challenge is the physiology of SGLTs in diverse locations throughout the body. For example, 1) in
the brain we need to determine why SGLTs are functionally expressed in the hippocampus, cerebral cortex, and
hypothalamus (258). Surely it is not to provide glucose to
neurons as they are amply endowed with GLUTs. Are the
SGLTs acting as glucose sensors or water channels in
these cells of the brain? 2) There are large gaps in our
understanding of the mechanism of milk secretion in the
lactating mammary gland. Are the SGLTs expressed in the
gland on the basolateral membrane to provide glucose
and galactose for lactose synthesis, or are the SGLTs
important for water secretion? 3) Similar questions arise
about the role of SGLTs in glucose secretion and absorption in salivary glands and interhepatic bile ducts. 4) What
is the significance of reported SGLT gene expression in
the heart, lung, muscle, testis, ovary, pancreas, and prostate? These questions are not just of physiological significance but are of vital interest in the pathophysiology of
glucose homeostasis and the upcoming introduction of
SGLT2 inhibitors to control blood glucose levels in diabetics.
It seems to us that we have just addressed the tip of
the iceberg about the biology of human SGLTs, and we
look forward to the entry of new investigators into this
field.
NOTE ADDED IN PROOF
Further insight into the transport mechanism comes
from our recent study using molecular dynamics, biochemistry, and a second crystal structure of vSGLT (238).
In essence, it is postulated that the conformational
change responsible for the transition from the outward to
inward occluded states (C3 to C4) is a rigid body movement where the “hash” (TM3, 4, 8, and 9) and “sugar”
bundles (TM 2, 6, and 7) rotate by 3° in the opposite
direction. This results in a displacement of TM8 by 4 Å,
and along with 13° kink in the inner half of TM1, that
loosens the Na2 site such that Na⫹ leaves to the cytoplasm within 9 ns. This exit causes a disruption in the
H-bond between Try-263 and Asn-64 permitting the Try263 side chain to assume a transient rotomer position
within 50 –100 ns allowing sugar to escape into the internal hydrophilic vestibule. Moreover, the rigid body movements of the hash and sugar bundles increase the volume
of the vestibule by 1,400 Å3, allowing free exit of the sugar
to the cytoplasm.
91 • APRIL 2011 •
www.prv.org
788
WRIGHT, LOO, AND HIRAYAMA
ACKNOWLEDGMENTS
Virtually all of the advances we have made in understanding SGLT were done in the Department of Physiology at UCLA
as a result of collaborative efforts by a team of talented people.
We are indebted to those students, fellows, and colleagues in the
team for their creative hard work. We also gratefully acknowledge the continuing collaborations with Jeff Abramson and his
group on the structure of SGLTs, and with colleagues in other
departments at UCLA, particularly Jorge Barrio in Medical and
Molecular Pharmacology, Martin Martin in Pediatrics, and Guido
Zampighi in Neurobiology. Thanks also to Thomas Zeuthen of
the University of Copenhagen for his insights into water trasport. We are thankful to Charles Hummel for his critical comments and to Wendy Ravenhill for drawing many of the figures
and her assistance in preparing the manuscript.
This work is in honor of R. K. Crane, 12/20/1919 –11/31/
2010.
Address for reprint requests and other correspondence:
E. M. Wright, Dept. of Physiology, David Geffen School of Medicine at UCLA, 10833 LeConte Ave., Los Angeles, CA 90095-1751
(e-mail: ewright@mednet.ucla.edu).
GRANTS
This work has been supported by various grants from the
National Institutes of Health (National Institute of Diabetes and
Digestive and Kidney Diseases) over the past 30 years, including
DK-19567.
DISCLOSURES
No conflicts of interest, financial or otherwise, are declared
by the authors.
REFERENCES
1. Abramson J, Wright EM. Structure and function of Na⫹-symporters with inverted repeats. Curr Opin Struct Biol 19: 425– 432, 2009.
2. Alvarado F, Crane RK. Studies on the mechanism of intestinal
absorption of sugars. VII. Phenylglycoside transport and its possible relationship to phlorizin inhibition of the active transport of
sugars by the small intestine. Biochim Biophys Acta 93: 116 –135,
1964.
3. Aouameur R, Da Cal S, Bissonnette P, Coady MJ, Lapointe
JY. SMIT2 mediates all myo-inositol uptake in apical membranes of
rat small intestine. Am J Physiol Gastrointest Liver Physiol 293:
G1300 –G1307, 2007.
4. Balen D, Ljubojevic M, Breljak D, Brzica H, Zlender V, Koepsell H, Sabolic I. Revised immunolocalization of the Na⫹-D-glucose cotransporter SGLT1 in rat organs with an improved antibody.
Am J Physiol Cell Physiol 295: C475–C489, 2008.
5. Barfuss DW, Schafer JA. Differences in active and passive glucose transport along the proximal nephron. Am J Physiol Renal
Fluid Electrolyte Physiol 241: F322–F332, 1981.
6. Barry RJ, Dikstein S, Matthews J, Smyth DH, Wright EM.
Electrical potentials associated with intestinal sugar transfer. J
Physiol 171: 316 –338, 1964.
7. Bessell EM, Foster AB, Westwood JH. The use of deoxyfluoroD-glucopyranoses and related compounds in a study of yeast
hexokinase specificity. Biochem J 128: 199 –204, 1972.
8. Bezanilla F, Perozo E, Papazian DM, Stefani E. Molecular basis
of gating charge immobilization in Shaker potassium channels.
Science 254: 679 – 683, 1991.
9. Bianchi L, Diez-Sampedro A. A single amino acid change converts the sugar sensor SGLT3 into a sugar transporter. PLoS One 5:
e10241.
Physiol Rev • VOL
10. Birnir B, Lee HS, Hediger MA, Wright EM. Expression and
characterization of the intestinal Na⫹/glucose cotransporter in
COS-7 cells. Biochim Biophys Acta 1048: 100 –104, 1990.
11. Birnir B, Loo DDF, Wright EM. Voltage-clamp studies of the
Na⫹/glucose cotransporter cloned from rabbit small intestine.
Pflügers Arch 418: 79 – 85, 1991.
12. Bormans GM, Van Oosterwyck G, De Groot TJ, Veyhl M,
Mortelmans L, Verbruggen AM, Koepsell H. Synthesis and biologic evaluation of (11)c-methyl-D-glucoside, a tracer of the sodium-dependent glucose transporters. J Nucl Med 44: 1075–1081,
2003.
13. Calado J, Sznajer Y, Metzger D, Rita A, Hogan MC, Kattamis
A, Scharf M, Tasic V, Greil J, Brinkert F, Kemper MJ, Santer
R. Twenty-one additional cases of familial renal glucosuria: absence of genetic heterogeneity, high prevalence of private mutations and further evidence of volume depletion. Nephrol Dial
Transplant 23: 3874 –3879, 2008.
14. Casneuf VF, Fonteyne P, Van Damme N, Demetter P, Pauwels
P, de Hemptinne B, De Vos M, Van de Wiele C, Peeters M.
Expression of SGLT1, Bcl-2 and p53 in primary pancreatic cancer
related to survival. Cancer Invest 26: 852– 859, 2008.
15. Charron FM, Blanchard MG, Lapointe JY. Intracellular hypertonicity is responsible for water flux associated with Na⫹/glucose
cotransport. Biophys J 90: 3546 –3554, 2006.
16. Chasis H, Jolliffe N, Smith HW. The action of phlorizin on the
excretion of glucose, xylose, sucrose, creatinine and urea by man.
J Clin Invest 12: 1083–1090, 1933.
17. Chen XZ, Coady MJ, Jackson F, Berteloot A, Lapointe JY.
Thermodynamic determination of the Na⫹:glucose coupling ratio
for the human SGLT1 cotransporter. Biophys J 69: 2405–2414,
1995.
18. Chen XZ, Coady MJ, Lapointe JY. Fast voltage clamp discloses
a new component of presteady-state currents from the Na⫹-glucose
cotransporter. Biophys J 71: 2544 –2552, 1996.
19. Chiorean EG, Dragovich T, Hamm J, Barrios CH, Gorini CF,
Langmuir VK, Kroll S, Jung DT, Tidmarsh GT, Loehrer PJ. A
Phase 2 trial of glufosfamide in combination with gemcitabine in
chemotherapy-naive pancreatic adenocarcinoma. Am J Clin Oncol
33: 111–116, 2010.
20. Choe S, Rosenberg JM, Abramson J, Wright EM, Grabe M. Water
permeation through the sodium-dependent galactose cotransporter
vSGLT. Biophys J 99: 56 –58, 2010.
21. Coady MJ, Wallendorff B, Gagnon DG, Lapointe JY. Identification of a novel Na⫹/myo-inositol cotransporter. J Biol Chem 277:
35219 –35224, 2002.
22. Cramer SC, Pardridge WM, Hirayama BA, Wright EM. Colocalization of GLUT2 glucose transporter, sodium/glucose cotransporter, and gamma-glutamyl transpeptidase in rat kidney with double-peroxidase immunocytochemistry. Diabetes 41: 766 –770, 1992.
23. Crane RK, Miller D, Bihler I. The restrictions on possible mechanisms of intestinal active transport of sugars. In: Membrane
Transport and Metabolism. Prague: Academic, 1960.
24. Crane RK. The gradient hypothesis and other models of carriermediated active transport. Rev Physiol Biochem Pharmacol 78:
99 –159, 1977.
25. Crane RK. Questions. In: Ion Gradient-Coupled Transport, edited
by Alvarado F, Van Os CH. New York: Elsevier, 1986, p. 431– 438.
26. Crane RK. The road to ion-coupled membrane processes. In:
Comprehensive Biochemistry, edited by Neuberger A, Van Deenen
LLM, Semenza G. New York: Elsevier Science, 1983, p. 43– 69.
27. Crane RK. Na⫹-dependent transport in the intestine and other
annual tissues. Federation, Proc 24: 1000 –1006, 1965.
28. Dahlin A, Royall J, Hohmann JG, Wang J. Expression profiling
of the solute carrier gene family in the mouse brain. J Pharmacol
Exp Ther 329: 558 –570, 2009.
29. Dai G, Levy O, Carrasco N. Cloning and characterization of the
thyroid iodide transporter. Nature 379: 458 – 460, 1996.
30. De Groot TJ, Veyhl M, Terwinghe C, Vanden Bempt V, Dupont
P, Mortelmans L, Verbruggen AM, Bormans GM, Koepsell H.
Synthesis of 18F-fluoroalkyl-beta-D-glucosides and their evaluation
as tracers for sodium-dependent glucose transporters. J Nucl Med
44: 1973–1981, 2003.
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
31. Debnam ES, Levin RJ. An experimental method of identifying
and quantifying the active transfer electrogenic component from
the diffusive component during sugar absorption measured in vivo.
J Physiol 246: 181–196, 1975.
32. Diedrich DF. The comparative effects of some phlorizin analogs
on the renal reabsorption of glucose. Biochim Biophys Acta 71:
688 –700, 1963.
33. Diez-Sampedro A. Involvement of amino acid 36 in TM1 in voltage
sensitivity in mouse Na⫹/glucose cotransporter SGLT1. J Membr
Biol 227: 57– 66, 2009.
34. Diez-Sampedro A, Eskandari S, Wright EM, Hirayama BA.
Na⫹-to-sugar stoichiometry of SGLT3. Am J Physiol Renal Physiol
280: F278 –F282, 2001.
35. Diez-Sampedro A, Hirayama BA, Osswald C, Gorboulev V,
Baumgarten K, Volk C, Wright EM, Koepsell H. A glucose
sensor hiding in a family of transporters. Proc Natl Acad Sci USA
100: 11753–11758, 2003.
36. Diez-Sampedro A, Loo DDF, Wright EM, Zampighi GA, Hirayama BA. Coupled sodium/glucose cotransport by SGLT1 requires a negative charge at position 454. Biochemistry 43: 13175–
13184, 2004.
37. Diez-Sampedro A, Lostao MP, Wright EM, Hirayama BA. Glycoside binding and translocation in Na⫹-dependent glucose
cotransporters: comparison of SGLT1 and SGLT3. J Membr Biol
176: 111–117, 2000.
38. Diez-Sampedro A, Wright EM, Hirayama BA. Residue 457 controls sugar binding and transport in the Na⫹/glucose cotransporter.
J Biol Chem 276: 49188 – 49194, 2001.
39. Dunham I, Shimizu N, Roe BA, Chissoe S, Hunt AR, Collins
JE, Bruskiewich R, Beare DM, Clamp M, Smink LJ, Ainscough
R, Almeida JP, Babbage A, Bagguley C, Bailey J, Barlow K,
Bates KN, Beasley O, Bird CP, Blakey S, Bridgeman AM, Buck
D, Burgess J, Burrill WD, O’Brien KP. The DNA sequence of
human chromosome 22. Nature 402: 489 – 495, 1999.
40. Dyer J, Daly K, Salmon KS, Arora DK, Kokrashvili Z, Margolskee RF, Shirazi-Beechey SP. Intestinal glucose sensing and
regulation of intestinal glucose absorption. Biochem Soc Trans 35:
1191–1194, 2007.
41. Dyer J, Vayro S, King TP, Shirazi-Beechey SP. Glucose sensing
in the intestinal epithelium. Eur J Biochem 270: 3377–3388, 2003.
42. Dyer J, Vayro S, Shirazi-Beechey SP. Mechanism of glucose
sensing in the small intestine. Biochem Soc Trans 31: 1140 –1142,
2003.
43. Dyer J, Wood IS, Palejwala A, Ellis A, Shirazi-Beechey SP.
Expression of monosaccharide transporters in intestine of diabetic
humans. Am J Physiol Gastrointest Liver Physiol 282: G241–G248,
2002.
44. Ehrenkranz JR, Lewis NG, Kahn CR, Roth J. Phlorizin: a review. Diabetes Metab Res Rev 21: 31–38, 2005.
45. Elfeber K, Kohler A, Lutzenburg M, Osswald C, Galla HJ,
Witte OW, Koepsell H. Localization of the Na⫹-D-glucose cotransporter SGLT1 in the blood-brain barrier. Histochem Cell Biol 121:
201–207, 2004.
46. Eskandari S, Wright EM, Kreman M, Starace DM, Zampighi
GA. Structural analysis of cloned plasma membrane proteins by
freeze-fracture electron microscopy. Proc Natl Acad Sci USA 95:
11235–11240, 1998.
47. Eskandari S, Wright EM, Loo DDF. Kinetics of the reverse mode
of the Na⫹/glucose cotransporter. J Membr Biol 204: 23–32, 2005.
48. Faham S, Watanabe A, Besserer GM, Cascio D, Specht A,
Hirayama BA, Wright EM, Abramson J. The crystal structure of
a sodium galactose transporter reveals mechanistic insights into
Na⫹/sugar symport. Science 321: 810 – 814, 2008.
49. Falk S, Guay A, Chenu C, Patil SD, Berteloot A. Reduction of
an eight-state mechanism of cotransport to a six-state model using
a new computer program. Biophys J 74: 816 – 830, 1998.
50. Fang Y, Jayaram H, Shane T, Kolmakova-Partensky L, Wu F,
Williams C, Xiong Y, Miller C. Structure of a prokaryotic virtual
proton pump at 3.2 A resolution. Nature 460: 1040 –1043, 2009.
51. Ferrannini E, Jimenez Ramos S, Salsali A, Tang W, List JF.
Dapagliflozin monotherapy in type 2 diabetic patients with inadequate glycemic control by diet and exercise: a randomized, doubleblind, placebo-controlled, phase III trial. Diabetes Care. In press.
Physiol Rev • VOL
789
52. Fuhrmann GF, Dernedde S, Frenking G. Phloretin keto-enol
tautomerism and inhibition of glucose transport in human erythrocytes (including effects of phloretin on anion transport). Biochim
Biophys Acta 1110: 105–111, 1992.
53. Gagnon DG, Holt A, Bourgeois F, Wallendorff B, Coady MJ,
Lapointe JY. Membrane topology of loop 13–14 of the Na⫹/glucose cotransporter (SGLT1): a SCAM and fluorescent labelling
study. Biochim Biophys Acta 1712: 173–184, 2005.
54. Gao X, Lu F, Zhou L, Dang S, Sun L, Li X, Wang J, Shi Y.
Structure and mechanism of an amino acid antiporter. Science 324:
1565–1568, 2009.
55. Gok F, Aydin HI, Kurt I, Gokcay E, Maeda M, Kasahara M. A
novel mutation of Na⫹/glucose cotransporter in a Turkish newborn
with congenital glucose-galactose malabsorption. J Pediatr Gastroenterol Nutr 40: 508 –511, 2005.
56. Gonzalez JA, Reimann F, Burdakov D. Dissociation between
sensing and metabolism of glucose in sugar sensing neurones. J
Physiol 587: 41– 48, 2009.
57. Gribble FM, Williams L, Simpson AK, Reimann F. A novel
glucose-sensing mechanism contributing to glucagon-like peptide-1
secretion from the GLUTag cell line. Diabetes 52: 1147–1154, 2003.
58. Hager K, Hazama A, Kwon HM, Loo DDF, Handler JS, Wright
EM. Kinetics and specificity of the renal Na⫹/myo-inositol cotransporter expressed in Xenopus oocytes. J Membr Biol 143: 103–113,
1995.
59. Hazama A, Loo DDF, Wright EM. Presteady-state currents of the
rabbit Na⫹/glucose cotransporter (SGLT1). J Membr Biol 155: 175–
186, 1997.
60. Hediger MA, Budarf ML, Emanuel BS, Mohandas TK, Wright
EM. Assignment of the human intestinal Na⫹/glucose cotransporter gene (SGLT1) to the q11.2-qter region of chromosome 22.
Genomics 4: 297–300, 1989.
61. Hediger MA, Coady MJ, Ikeda TS, Wright EM. Expression
cloning and cDNA sequencing of the Na⫹/glucose co-transporter.
Nature 330: 379 –381, 1987.
62. Hediger MA, Ikeda T, Coady M, Gundersen CB, Wright EM.
Expression of size-selected mRNA encoding the intestinal Na/glucose cotransporter in Xenopus laevis oocytes. Proc Natl Acad Sci
USA 84: 2634 –2637, 1987.
63. Hediger MA, Mendlein J, Lee HS, Wright EM. Biosynthesis of
the cloned intestinal Na⫹/glucose cotransporter. Biochim Biophys
Acta 1064: 360 –364, 1991.
64. Hediger MA, Turk E, Wright EM. Homology of the human intestinal Na⫹/glucose and Escherichia coli Na⫹/proline cotransporters.
Proc Natl Acad Sci USA 86: 5748 –5752, 1989.
65. Helmke BM, Reisser C, Idzko M, Dyckhoff G, Herold-Mende
C. Expression of SGLT-1 in preneoplastic and neoplastic lesions of
the head and neck. Oral Oncol 40: 28 –35, 2004.
66. Hirayama BA, Diez-Sampedro A, Wright EM. Common mechanisms of inhibition for the Na⫹/glucose (hSGLT1) and Na⫹/Cl⫺/
GABA (hGAT1) cotransporters. Br J Pharmacol 134: 484 – 495,
2001.
67. Hirayama BA, Loo DDF, Diez-Sampedro A, Leung DW,
Meinild AK, Lai-Bing M, Turk E, Wright EM. Sodium-dependent
reorganization of the sugar-binding site of SGLT1. Biochemistry
46: 13391–13406, 2007.
68. Hirayama BA, Loo DDF, Wright EM. Cation effects on protein
conformation and transport in the Na⫹/glucose cotransporter. J
Biol Chem 272: 2110 –2115, 1997.
69. Hirayama BA, Lostao MP, Panayotova-Heiermann M, Loo DD,
Turk E, Wright EM. Kinetic and specificity differences between
rat, human, and rabbit Na⫹-glucose cotransporters (SGLT-1). Am J
Physiol Gastrointest Liver Physiol 270: G919 –G926, 1996.
70. Hirayama BA, Wong HC, Smith CD, Hagenbuch BA, Hediger
MA, Wright EM. Intestinal and renal Na⫹/glucose cotransporters
share common structures. Am J Physiol Cell Physiol 261: C296 –
C304, 1991.
71. Hirayama BA, Wright EM. Glycosylation of the rabbit intestinal
brush border Na⫹/glucose cotransporter. Biochim Biophys Acta
1103: 37– 44, 1992.
72. Hirsch JR, Loo DDF, Wright EM. Regulation of Na⫹/glucose
cotransporter expression by protein kinases in Xenopus laevis
oocytes. J Biol Chem 271: 14740 –14746, 1996.
91 • APRIL 2011 •
www.prv.org
790
WRIGHT, LOO, AND HIRAYAMA
73. Hirschhorn N, Greenough WB 3rd. Progress in oral rehydration
therapy. Sci Am 264: 50 –56, 1991.
74. Hirschhorn N, Kinzie JL, Sachar DB, Northrup RS, Taylor JO,
Ahmad SZ, Phillips RA. Decrease in net stool output in cholera
during intestinal perfusion with glucose-containing solutions. N
Engl J Med 279: 176 –181, 1968.
75. Hopfer U. Membrane transport mechanisms for hexoses and
amino acids in the small intestine. In: Physiology of the Gastrointestinal Tract (2nd ed.), edited by Johnson LR, Christensen J,
Jackson MJ, Jacobson ED, Walsh JH. New York: Raven, 1987, p.
1499 –1526.
76. Hopfer U, Nelson K, Perrotto J, Isselbacher KJ. Glucose transport in isolated brush border membrane from rat small intestine. J
Biol Chem 248: 25–32, 1973.
77. Horiba N, Masuda S, Ohnishi C, Takeuchi D, Okuda M, Inui K.
Na⫹-dependent fructose transport via rNaGLT1 in rat kidney. FEBS
Lett 546: 276 –280, 2003.
78. Horiba N, Masuda S, Takeuchi A, Takeuchi D, Okuda M, Inui
K. Cloning and characterization of a novel Na⫹-dependent glucose
transporter (NaGLT1) in rat kidney. J Biol Chem 278: 14669 –14676,
2003.
79. Hosang M, Gibbs EM, Diedrich DF, Semenza G. Photoaffinity
labeling and identification of (a component of) the small-intestinal
Na⫹,D-glucose transporter using 4-azidophlorizin. FEBS Lett 130:
244 –248, 1981.
80. Huang SC, Truong D, Wu HM, Chatziioannou AF, Shao W, Wu
AM, Phelps ME. An internet-based “kinetic imaging system” (KIS)
for MicroPET. Mol Imaging Biol 7: 330 –341, 2005.
81. Hummel C, Lu C, Loo DDF, Hirayama BA, Voss AA, Wright
EM. Glucose transport by human renal Na⫹/D-glucose cotransporters SGLT1 and SGLT2. Am J Physiol Cell Physiol 300: C14 –C21,
2011.
82. Ikeda TS, Hwang ES, Coady MJ, Hirayama BA, Hediger MA,
Wright EM. Characterization of a Na⫹/glucose cotransporter
cloned from rabbit small intestine. J Membr Biol 110: 87–95, 1989.
83. Isaji M. Sodium-glucose cotransporter inhibitors for diabetes.
Curr Opin Invest Drugs 8: 285–292, 2007.
84. Ishikawa N, Oguri T, Isobe T, Fujitaka K, Kohno N. SGLT gene
expression in primary lung cancers and their metastatic lesions.
Jpn J Cancer Res 92: 874 – 879, 2001.
85. Kaback HR, Stadtman ER. Proline uptake by an isolated cytoplasmic membrane preparation of Escherichia coli. Proc Natl Acad
Sci USA 55: 920 –927, 1966.
86. Kanai Y, Lee WS, You G, Brown D, Hediger MA. The human
kidney low affinity Na⫹/glucose cotransporter SGLT2. Delineation
of the major renal reabsorptive mechanism for D-glucose. J Clin
Invest 93: 397– 404, 1994.
87. Kasahara M, Maeda M, Hayashi S, Mori Y, Abe T. A missense
mutation in the Na⫹/glucose cotransporter gene SGLT1 in a patient
with congenital glucose-galactose malabsorption: normal trafficking but inactivation of the mutant protein. Biochim Biophys Acta
1536: 141–147, 2001.
88. Katsuno K, Fujimori Y, Takemura Y, Hiratochi M, Itoh F,
Komatsu Y, Fujikura H, Isaji M. Sergliflozin, a novel selective
inhibitor of low-affinity sodium glucose cotransporter (SGLT2),
validates the critical role of SGLT2 in renal glucose reabsorption
and modulates plasma glucose level. J Pharmacol Exp Ther 320:
323–330, 2007.
89. Kellett GL, Brot-Laroche E. Apical GLUT2: a major pathway of
intestinal sugar absorption. Diabetes 54: 3056 –3062, 2005.
90. Kellett GL, Helliwell PA. The diffusive component of intestinal
glucose absorption is mediated by the glucose-induced recruitment
of GLUT2 to the brush-border membrane. Biochem J 350: 155–162,
2000.
91. Kessler M, Semenza G. The small-intestinal Na⫹,D-glucose
cotransporter: an asymmetric gated channel (or pore) responsive
to delta psi. J Membr Biol 76: 27–56, 1983.
92. Kimmich GA. Intestinal absorption of sugar. In: Physiology of the
Gastrointestinal Tract, edited by Johnson LR, Christensen J,
Grossman MI, Jacobson ED, Schultz SG. New York: Raven, 1981, p.
1035–1061.
Physiol Rev • VOL
93. Kimmich GA, Randles J. Sodium-sugar coupling stoichiometry in
chick intestinal cells. Am J Physiol Cell Physiol 247: C74 –C82,
1984.
94. Kleinzellerr AKA. Membrane Transport and Metabolism. Praha:
Publishing House of the Czechoslovak Academy of Sciences, 1961,
p. 608.
95. Komoroski B, Vachharajani N, Boulton D, Kornhauser D, Geraldes M, Li L, Pfister M. Dapagliflozin, a novel SGLT2 inhibitor,
induces dose-dependent glucosuria in healthy subjects. Clin Pharmacol Ther 85: 520 –526, 2009.
96. Komoroski B, Vachharajani N, Feng Y, Li L, Kornhauser D,
Pfister M. Dapagliflozin, a novel, selective SGLT2 inhibitor, improved glycemic control over 2 weeks in patients with type 2
diabetes mellitus. Clin Pharmacol Ther 85: 513–519, 2009.
97. Kopito RR, Lodish HF. Primary structure and transmembrane
orientation of the murine anion exchange protein. Nature 316:
234 –238, 1985.
98. Korn T, Kuhlkamp T, Track C, Schatz I, Baumgarten K, Gorboulev V, Koepsell H. The plasma membrane-associated protein
RS1 decreases transcription of the transporter SGLT1 in confluent
LLC-PK1 cells. J Biol Chem 276: 45330 – 45340, 2001.
99. Krofchick D, Huntley SA, Silverman M. Transition states of the
high-affinity rabbit Na⫹/glucose cotransporter SGLT1 as determined from measurement and analysis of voltage-dependent
charge movements. Am J Physiol Cell Physiol 287: C46 –C54, 2004.
100. Krofchick D, Silverman M. Investigating the conformational
states of the rabbit Na⫹/glucose cotransporter. Biophys J 84: 3690 –
3702, 2003.
101. Kroiss M, Leyerer M, Gorboulev V, Kuhlkamp T, Kipp H,
Koepsell H. Transporter regulator RS1 (RSC1A1) coats the transGolgi network and migrates into the nucleus. Am J Physiol Renal
Physiol 291: F1201–F1212, 2006.
102. Kumar A, Tyagi NK, Arevalo E, Miller KW, Kinne RK. A
proteomic study of sodium/D-glucose cotransporter 1 (SGLT1):
topology of loop 13 and coverage of other functionally important
domains. Biochim Biophys Acta 1774: 968 –974, 2007.
103. Kunze WA, Furness JB. The enteric nervous system and regulation of intestinal motility. Annu Rev Physiol 61: 117–142, 1999.
104. Kwon HM, Yamauchi A, Uchida S, Preston AS, Garcia-Perez
A, Burg MB, Handler JS. Cloning of the cDNa for a Na⫹/myoinositol cotransporter, a hypertonicity stress protein. J Biol Chem
267: 6297– 6301, 1992.
105. Lam JT, Martin MG, Turk E, Hirayama BA, Bosshard NU,
Steinmann B, Wright EM. Missense mutations in SGLT1 cause
glucose-galactose malabsorption by trafficking defects. Biochim
Biophys Acta 1453: 297–303, 1999.
106. Lambotte S, Veyhl M, Kohler M, Morrison-Shetlar AI, Kinne
RK, Schmid M, Koepsell H. The human gene of a protein that
modifies Na⫹-D-glucose co-transport. DNA Cell Biol 15: 769 –777,
1996.
107. Landau BR, Bernstein L, Wilson TH. Hexose transport by hamster intestine in vitro. Am J Physiol 203: 237–240, 1962.
108. Laplane R, Polonovski C, Etienne M, Debray P, Lods JC,
Pissarro B. L’intolerance aux sucres a transfert intestinal actif.
Arch Pr Pediatr 19: 895–944, 1962.
109. Lapointe JY. Response to Zeuthen and Zeuthen’s comment to the
editor: enough local hypertonicity is enough. Biophys J 93: 1417–
1419, 2007.
110. Lauger P, Jauch P. Microscopic description of voltage effects on
ion-driven cotransport systems. J Membr Biol 91: 275–284, 1986.
111. Lazaridis KN, Pham L, Vroman B, de Groen PC, LaRusso NF.
Kinetic and molecular identification of sodium-dependent glucose
transporter in normal rat cholangiocytes. Am J Physiol Gastrointest Liver Physiol 272: G1168 –G1174, 1997.
112. Le Coutre J, Turk E, Kaback HR, Wright EM. Ligand-induced
differences in secondary structure of the Vibrio parahaemolyticus
Na⫹/galactose cotransporter. Biochemistry 41: 8082– 8086, 2002.
113. Leicht S, Grohmann S, Page K, Streicher R, Mark M, Eichelmann P. hSGLT5 (SLC5A10) is a sodium-dependent sugar transporter exclusively expressed in the kidney. Am Diabetes Assoc 66:
1512-P, 2006.
114. Lescale-Matys L, Dyer J, Scott D, Freeman TC, Wright EM,
Shirazi-Beechey SP. Regulation of the ovine intestinal Na⫹/glu-
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
115.
116.
117.
118.
119.
120.
121.
122.
123.
124.
125.
126.
127.
128.
129.
130.
131.
132.
133.
134.
135.
136.
137.
cose co-transporter (SGLT1) is dissociated from mRNA abundance.
Biochem J 291: 435– 440, 1993.
Leung DW, Loo DDF, Hirayama BA, Zeuthen T, Wright EM.
Urea transport by cotransporters. J Physiol 528: 251–257, 2000.
Leung DW, Turk E, Kim O, Wright EM. Functional expression of
the Vibrio parahaemolyticus Na⫹/galactose (vSGLT) cotransporter in Xenopus laevis oocytes. J Membr Biol 187: 65–70, 2002.
Li J, Tajkhorshid E. Ion-releasing state of a secondary membrane
transporter. Biophys J 97: 29 –31, 2009.
Lin X, Ma L, Fitzgerald RL, Ostlund RE Jr. Human sodium/
inositol cotransporter 2 (SMIT2) transports inositols but not glucose in L6 cells. Arch Biochem Biophys 481: 197–201, 2009.
Lindquist B, Meeuwisse GW. Chronic diarrhoea caused by monosaccharide malabsorption. Acta Paediatr 51: 674 – 685, 1962.
List JF, Woo V, Morales E, Tang W, Fiedorek FT. Sodiumglucose cotransport inhibition with dapagliflozin in type 2 diabetes.
Diabetes Care 32: 650 – 657, 2009.
Liu M, Seino S, Kirchgessner AL. Identification and characterization of glucoresponsive neurons in the enteric nervous system. J
Neurosci 19: 10305–10317, 1999.
Liu T, Speight P, Silverman M. Reanalysis of structure/function
correlations in the region of transmembrane segments 4 and 5 of
the rabbit sodium/glucose cotransporter. Biochem Biophys Res
Commun 378: 133–138, 2009.
Lo B, Silverman M. Cysteine scanning mutagenesis of the segment between putative transmembrane helices IV and V of the high
affinity Na⫹/glucose cotransporter SGLT1. Evidence that this region participates in the Na⫹ and voltage dependence of the transporter. J Biol Chem 273: 29341–29351, 1998.
Loike JD, Hickman S, Kuang K, Xu M, Cao L, Vera JC, Silverstein SC, Fischbarg J. Sodium-glucose cotransporters display
sodium- and phlorizin-dependent water permeability. Am J Physiol
Cell Physiol 271: C1774 –C1779, 1996.
Loo DDF, Hirayama BA, Sala-Rabanal M, Wright EM. How
drugs interact with transporters; SGLT1 as a model. J Membr Biol
223: 87–106, 2008.
Loo DDF, Eskandari S, Boorer KJ, Sarkar HK, Wright EM.
Role of Cl⫺ in electrogenic Na⫹-coupled cotransporters GAT1 and
SGLT1. J Biol Chem 275: 37414 –37422, 2000.
Loo DDF, Hazama A, Supplisson S, Turk E, Wright EM. Relaxation kinetics of the Na⫹/glucose cotransporter. Proc Natl Acad
Sci USA 90: 5767–5771, 1993.
Loo DDF, Hirayama BA, Cha A, Bezanilla F, Wright EM.
Perturbation analysis of the voltage-sensitive conformational
changes of the Na⫹/glucose cotransporter. J Gen Physiol 125:
13–36, 2005.
Loo DDF, Hirayama BA, Gallardo EM, Lam JT, Turk E, Wright
EM. Conformational changes couple Na⫹ and glucose transport.
Proc Natl Acad Sci USA 95: 7789 –7794, 1998.
Loo DDF, Hirayama BA, Karakossian MH, Meinild AK, Wright
EM. Conformational dynamics of hSGLT1 during Na⫹/glucose
cotransport. J Gen Physiol 128: 701–720, 2006.
Loo DDF, Hirayama BA, Meinild AK, Chandy G, Zeuthen T,
Wright EM. Passive water and ion transport by cotransporters. J
Physiol 518: 195–202, 1999.
Loo DDF, Hirayama BA, Sala-Rabanal M, Wright EM. How
drugs interact with transporters: SGLT1 as a model. J Membr Biol
223: 87–106, 2008.
Loo DDF, Wright EM, Zeuthen T. Water pumps. J Physiol 542:
53– 60, 2002.
Loo DDF, Zeuthen T, Chandy G, Wright EM. Cotransport of
water by the Na⫹/glucose cotransporter. Proc Natl Acad Sci USA
93: 13367–13370, 1996.
Loo DDF, Hirayama BA, Karakossian MH, Meinmild AK,
Wright EM. Conformational dynamics of hSGLT1 during Na⫹/
glucose cotransport. J Gen Physiol 128: 701–720, 2006.
Lostao MP, Hirayama BA, Loo DDF, Wright EM. Phenylglucosides and the Na⫹/glucose cotransporter (SGLT1): analysis of interactions. J Membr Biol 142: 161–170, 1994.
Lostao MP, Hirayama BA, Panayotova-Heiermann M, Sampogna SL, Bok D, Wright EM. Arginine-427 in the Na⫹/glucose
cotransporter (SGLT1) is involved in trafficking to the plasma
membrane. FEBS Lett 377: 181–184, 1995.
Physiol Rev • VOL
791
138. Mackenzie B, Loo DDF, Panayotova-Heiermann M, Wright
EM. Biophysical characteristics of the pig kidney Na⫹/glucose
cotransporter SGLT2 reveal a common mechanism for SGLT1 and
SGLT2. J Biol Chem 271: 32678 –32683, 1996.
139. Mackenzie B, Loo DDF, Wright EM. Relationships between Na⫹/
glucose cotransporter (SGLT1) currents and fluxes. J Membr Biol
162: 101–106, 1998.
140. Martin MG, Lostao MP, Turk E, Lam J, Kreman M, Wright
EM. Compound missense mutations in the sodium/D-glucose
cotransporter result in trafficking defects. Gastroenterology 112:
1206 –1212, 1997.
141. Martin MG, Turk E, Kerner C, Zabel B, Wirth S, Wright EM.
Prenatal identification of a heterozygous status in two fetuses at
risk for glucose-galactose malabsorption. Prenat Diagn 16: 458 –
462, 1996.
142. Martin MG, Turk E, Lostao MP, Kerner C, Wright EM. Defects
in Na⫹/glucose cotransporter (SGLT1) trafficking and function
cause glucose-galactose malabsorption. Nat Genet 12: 216 –220,
1996.
143. Martin MG, Wang J, Solorzano-Vargas RS, Lam JT, Turk E,
Wright EM. Regulation of the human Na⫹-glucose cotransporter
gene, SGLT1, by HNF-1 and Sp1. Am J Physiol Gastrointest Liver
Physiol 278: G591–G603, 2000.
144. Masyuk AI, Masyuk TV, Tietz PS, Splinter PL, LaRusso NF.
Intrahepatic bile ducts transport water in response to absorbed
glucose. Am J Physiol Cell Physiol 283: C785–C791, 2002.
145. Meinild A, Klaerke DA, Loo DDF, Wright EM, Zeuthen T. The
human Na⫹-glucose cotransporter is a molecular water pump. J
Physiol 508: 15–21, 1998.
146. Meinild AK, Hirayama BA, Wright EM, Loo DDF. Fluorescence
studies of ligand-induced conformational changes of the Na⫹/glucose cotransporter. Biochemistry 41: 1250 –1258, 2002.
147. Meng W, Ellsworth BA, Nirschl AA, McCann PJ, Patel M,
Girotra RN, Wu G, Sher PM, Morrison EP, Biller SA, Zahler R,
Deshpande PP, Pullockaran A, Hagan DL, Morgan N, Taylor
JR, Obermeier MT, Humphreys WG, Khanna A, Discenza L,
Robertson JG, Wang A, Han S, Wetterau JR, Janovitz EB,
Flint OP, Whaley JM, Washburn WN. Discovery of dapagliflozin:
a potent, selective renal sodium-dependent glucose cotransporter 2
(SGLT2) inhibitor for the treatment of type 2 diabetes. J Med Chem
51: 1145–1149, 2008.
148. Mitchell P. Molecule, group and electron translocation through
natural membrane. In: The Structure of Function of the Membrane,
edited by Grant JK, Bell DJ. London: Biochemical Society, 1963,
Symposium 22, p. 142–168.
149. Mizuma T, Ohta K, Hayashi M, Awazu S. Intestinal active absorption of sugar-conjugated compounds by glucose transport system: implication of improvement of poorly absorbable drugs.
Biochem Pharmacol 43: 2037–2039, 1992.
150. Mueckler M, Caruso C, Baldwin SA, Panico M, Blench I,
Morris HR, Allard WJ, Lienhard GE, Lodish HF. Sequence and
structure of a human glucose transporter. Science 229: 941–945,
1985.
151. Nishimura M, Naito S. Tissue-specific mRNA expression profiles
of human ATP-binding cassette and solute carrier transporter superfamilies. Drug Metab Pharmacokinet 20: 452– 477, 2005.
152. Noskov SY, Roux B. Control of ion selectivity in LeuT: two Na⫹
binding sites with two different mechanisms. J Mol Biol 377: 804 –
818, 2008.
154. O’Malley D, Reimann F, Simpson AK, Gribble FM. Sodiumcoupled glucose cotransporters contribute to hypothalamic glucose sensing. Diabetes 55: 3381–3386, 2006.
155. Obermeier S, Huselweh B, Tinel H, Kinne RH, Kunz C. Expression of glucose transporters in lactating human mammary
gland epithelial cells. Eur J Nutr 39: 194 –200, 2000.
156. Oku A, Ueta K, Arakawa K, Ishihara T, Nawano M, Kuronuma
Y, Matsumoto M, Saito A, Tsujihara K, Anai M, Asano T,
Kanai Y, Endou H. T-1095, an inhibitor of renal Na⫹-glucose
cotransporters, may provide a novel approach to treating diabetes.
Diabetes 48: 1794 –1800, 1999.
157. Page MJ, Carrell CJ, Di Cera E. Engineering protein allostery:
1.05 A resolution structure and enzymatic properties of a Na⫹activated trypsin. J Mol Biol 378: 666 – 672, 2008.
91 • APRIL 2011 •
www.prv.org
792
WRIGHT, LOO, AND HIRAYAMA
158. Page MJ, Di Cera E. Role of Na⫹ and K⫹ in enzyme function.
Physiol Rev 86: 1049 –1092, 2006.
159. Pajor AM, Randolph KM, Kerner SA, Smith CD. Inhibitor binding in the human renal low- and high-affinity Na⫹/glucose cotransporters. J Pharmacol Exp Ther 324: 985–991, 2008.
160. Palazzo M, Gariboldi S, Zanobbio L, Selleri S, Dusio GF,
Mauro V, Rossini A, Balsari A, Rumio C. Sodium-dependent
glucose transporter-1 as a novel immunological player in the intestinal mucosa. J Immunol 181: 3126 –3136, 2008.
161. Panayotova-Heiermann M, Loo DDF, Kong CT, Lever JE,
Wright EM. Sugar binding to Na⫹/glucose cotransporters is determined by the carboxyl-terminal half of the protein. J Biol Chem
271: 10029 –10034, 1996.
162. Panayotova-Heiermann M, Loo DDF, Lam JT, Wright EM.
Neutralization of conservative charged transmembrane residues in
the Na⫹/glucose cotransporter SGLT1. Biochemistry 37: 10522–
10528, 1998.
163. Panayotova-Heiermann M, Loo DDF, Lostao MP, Wright EM.
Sodium/D-glucose cotransporter charge movements involve polar
residues. J Biol Chem 269: 21016 –21020, 1994.
164. Panayotova-Heiermann M, Loo DDF, Wright EM. Kinetics of
steady-state currents and charge movements associated with the
rat Na⫹/glucose cotransporter. J Biol Chem 270: 27099 –27105,
1995.
165. Panayotova-Heiermann M, Wright EM. Mapping the urea channel through the rabbit Na⫹-glucose cotransporter SGLT1. J Physiol
535: 419 – 425, 2001.
166. Parent L, Supplisson S, Loo DDF, Wright EM. Electrogenic
properties of the cloned Na⫹/glucose cotransporter. I. Voltageclamp studies. J Membr Biol 125: 49 – 62, 1992.
167. Parent L, Supplisson S, Loo DDF, Wright EM. Electrogenic
properties of the cloned Na⫹/glucose cotransporter. II. A transport
model under nonrapid equilibrium conditions. J Membr Biol 125:
63–79, 1992.
168. Parent L, Wright EM. Electrophysiology of the Na⫹/glucose
cotransporter. Soc Gen Physiol Ser 48: 263–281, 1993.
169. Peerce BE, Wright EM. Conformational changes in the intestinal
brush border sodium-glucose cotransporter labeled with fluorescein isothiocyanate. Proc Natl Acad Sci USA 81: 2223–2226, 1984.
170. Phelps ME. PET Molecular Imaging and It’s Biological Applications. New York: Springer, 2004.
171. Pierce NF, Banwell JG, Rupak DM, Mitra RC, Caranasos GJ,
Keimowitz RI, Mondal A, Manji PM. Effect of intragastric glucose-electrolyte infusion upon water and electrolyte balance in
Asiatic cholera. Gastroenterology 55: 333–343, 1968.
172. Pontoglio M, Prie D, Cheret C, Doyen A, Leroy C, Froguel P,
Velho G, Yaniv M, Friedlander G. HNF1alpha controls renal
glucose reabsorption in mouse and man. EMBO Rep 1: 359 –365,
2000.
173. Poppe R, Karbach U, Gambaryan S, Wiesinger H, Lutzenburg
M, Kraemer M, Witte OW, Koepsell H. Expression of the Na⫹D-glucose cotransporter SGLT1 in neurons. J Neurochem 69: 84 –94,
1997.
174. Prasad PD, Wang H, Kekuda R, Fujita T, Fei YJ, Devoe LD,
Leibach FH, Ganapathy V. Cloning and functional expression of
a cDNA encoding a mammalian sodium-dependent vitamin transporter mediating the uptake of pantothenate, biotin, and lipoate. J
Biol Chem 273: 7501–7506, 1998.
175. Quick M, Loo DDF, Wright EM. Neutralization of a conserved
amino acid residue in the human Na⫹/glucose transporter
(hSGLT1) generates a glucose-gated H⫹ channel. J Biol Chem 276:
1728 –1734, 2001.
176. Quick M, Tomasevic J, Wright EM. Functional asymmetry of the
human Na⫹/glucose transporter (hSGLT1) in bacterial membrane
vesicles. Biochemistry 42: 9147–9152, 2003.
177. Quick M, Wright EM. Employing Escherichia coli to functionally
express, purify, and characterize a human transporter. Proc Natl
Acad Sci USA 99: 8597– 8601, 2002.
178. Quiocho FA. Protein-carbohydrate interactions: basic molecular
features. Pure Appl Chem 61: 1293–1306, 1989.
179. Rao MC. Oral rehydration therapy: new explanations for an old
remedy. Annu Rev Physiol 66: 385– 417, 2004.
Physiol Rev • VOL
180. Raybould HE. Visceral perception: sensory transduction in visceral afferents and nutrients. Gut 51 Suppl 1: i11–i14, 2002.
181. Raybould HE, Zittel TT. Inhibition of gastric motility induced by
intestinal glucose in awake rats: role of Na⫹-glucose co-transporter. Neurogastroenterol Motil 7: 9 –14, 1995.
182. Reinhardt J, Veyhl M, Wagner K, Gambaryan S, Dekel C,
Akhoundova A, Korn T, Koepsell H. Cloning and characterization of the transport modifier RS1 from rabbit which was previously assumed to be specific for Na⫹-D-glucose cotransport.
Biochim Biophys Acta 1417: 131–143, 1999.
183. Ressl S, Terwisscha van Scheltinga AC, Vonrhein C, Ott V,
Ziegler C. Molecular basis of transport and regulation in the
Na⫹/betaine symporter BetP. Nature 458: 47–52, 2009.
184. Robinson J. Moving Question: A History of Membrane Transport
and Bioenergetics. Bethesda, MD: Am. Physiol. Soc., 1997.
185. Rossetti L, Smith D, Shulman GI, Papachristou D, DeFronzo
RA. Correction of hyperglycemia with phlorizin normalizes tissue
sensitivity to insulin in diabetic rats. J Clin Invest 79: 1510 –1515,
1987.
186. Ruxin JN. Magic bullet: the history of oral rehydration therapy.
Med Hist 38: 363–397, 1994.
187. Sabino-Silva R, Freitas HS, Lamers ML, Okamoto MM, Santos
MF, Machado UF. Na⫹-glucose cotransporter SGLT1 protein in
salivary glands: potential involvement in the diabetes-induced decrease in salivary flow. J Membr Biol 228: 63– 69, 2009.
188. Sala-Rabanal M, Loo DDF, Hirayama BA, Turk E, Wright EM.
Molecular interactions between dipeptides, drugs and the human
intestinal H⫹-oligopeptide cotransporter hPEPT1. J Physiol 574:
149 –166, 2006.
189. Santer R, Groth S, Kinner M, Dombrowski A, Berry GT, Brodehl J, Leonard JV, Moses S, Norgren S, Skovby F, Schneppenheim R, Steinmann B, Schaub J. The mutation spectrum of
the facilitative glucose transporter gene SLC2A2 (GLUT2) in patients with Fanconi-Bickel syndrome. Hum Genet 110: 21–29, 2002.
190. Santer R, Kinner M, Lassen CL, Schneppenheim R, Eggert P,
Bald M, Brodehl J, Daschner M, Ehrich JH, Kemper M, Li
Volti S, Neuhaus T, Skovby F, Swift PG, Schaub J, Klaerke D.
Molecular analysis of the SGLT2 gene in patients with renal glucosuria. J Am Soc Nephrol 14: 2873–2882, 2003.
191. Sauer GA, Nagel G, Koepsell H, Bamberg E, Hartung K. Voltage and substrate dependence of the inverse transport mode of the
rabbit Na⫹/glucose cotransporter (SGLT1). FEBS Lett 469: 98 –100,
2000.
192. Schmidt UM, Eddy B, Fraser CM, Venter JC, Semenza G.
Isolation of (a subunit of) the Na⫹/D-glucose cotransporter(s) of
rabbit intestinal brush border membranes using monoclonal antibodies. FEBS Lett 161: 279 –283, 1983.
193. Schneider AJ, Kinter WB, Stirling CE. Glucose-galactose malabsorption. Report of a case with autoradiographic studies of a
mucosal biopsy. N Engl J Med 274: 305–312, 1966.
194. Scholl-Burgi S, Santer R, Ehrich JH. Long-term outcome of
renal glucosuria type 0: the original patient and his natural history.
Nephrol Dial Transplant 19: 2394 –2396, 2004.
195. Schultz SG. Ion-coupled transport of organic solutes across biological membranes. In: Physiology of Membrane Disorders, edited
by Andreoli TE, Hoffman JF, Fanestil DD. 1985, New York: Plenum,
p. 283–294.
196. Schultz SG, Curran PF. Coupled transport of sodium and organic
solutes. Physiol Rev 50: 637–718, 1970.
197. Sharp PA, Debnam ES. The role of cAMP in the control of sugar
transport across the brush-border and basolateral membranes of
rat jejunal enterocytes. Exp Physiol 79: 203–214, 1994.
198. Shennan DB, Beechey RB. Mechanisms involved in the uptake of
D-glucose into the milk producing cells of rat mammary tissue.
Biochem Biophys Res Commun 211: 986 –990, 1995.
199. Shimamura T, Weyand S, Beckstein O, Rutherford NG, Hadden JM, Sharples D, Sansom MS, Iwata S, Henderson PJ,
Cameron AD. Molecular basis of alternating access membrane
transport by the sodium-hydantoin transporter Mhp1. Science 328:
470 – 473, 2010.
200. Shimizu T, Okamoto I, Tamura K, Satoh T, Miyazaki M,
Akashi Y, Ozaki T, Fukuoka M, Nakagawa K. Phase I clinical
and pharmacokinetic study of the glucose-conjugated cytotoxic
91 • APRIL 2011 •
www.prv.org
BIOLOGY OF HUMAN SODIUM GLUCOSE TRANSPORTERS
201.
202.
203.
204.
205.
206.
207.
208.
209.
210.
211.
212.
213.
214.
215.
216.
217.
218.
219.
agent D: ⫺19575 (glufosfamide) in patients with solid tumors.
Cancer Chemother Pharmacol 65: 243–250, 2010.
Shirazi-Beechey SP, Hirayama BA, Wang Y, Scott D, Smith
MW, Wright EM. Ontogenic development of lamb intestinal sodium-glucose co-transporter is regulated by diet. J Physiol 437: 699 –
708, 1991.
Smith CD, Hirayama BA, Wright EM. Baculovirus-mediated expression of the Na⫹/glucose cotransporter in Sf9 cells. Biochim
Biophys Acta 1104: 151–159, 1992.
Soylu OB, Ecevit C, Altinoz S, Ozturk AA, Temizkan AK,
Maeda M, Kasahara M. Nephrocalcinosis in glucose-galactose
malabsorption: nephrocalcinosis and proximal tubular dysfunction
in a young infant with a novel mutation of SGLT1. Eur J Pediatr
167: 1395–1398, 2008.
Stefani E, Bezanilla F. Cut-open oocyte voltage-clamp technique.
Methods Enzymol 293: 300 –318, 1998.
Stevens BR, Fernandez A, Hirayama B, Wright EM, Kempner
ES. Intestinal brush border membrane Na⫹/glucose cotransporter
functions in situ as a homotetramer. Proc Natl Acad Sci USA 87:
1456 –1460, 1990.
Stumpel F, Burcelin R, Jungermann K, Thorens B. Normal
kinetics of intestinal glucose absorption in the absence of GLUT2:
evidence for a transport pathway requiring glucose phosphorylation and transfer into the endoplasmic reticulum. Proc Natl Acad
Sci USA 98: 11330 –11335, 2001.
Suzuki T, Fujikura K, Koyama H, Matsuzaki T, Takahashi Y,
Takata K. The apical localization of SGLT1 glucose transporter is
determined by the short amino acid sequence in its N-terminal
domain. Eur J Cell Biol 80: 765–774, 2001.
Taglialatela M, Toro L, Stefani E. Novel voltage clamp to record
small, fast currents from ion channels expressed in Xenopus
oocytes. Biophys J 61: 78 – 82, 1992.
Takata K, Kasahara T, Kasahara M, Ezaki O, Hirano H. Immunohistochemical localization of Na⫹-dependent glucose transporter in rat jejunum. Cell Tissue Res 267: 3–9, 1992.
Takata K, Kasahara T, Kasahara M, Ezaki O, Hirano H. Localization of Na⫹-dependent active type and erythrocyte/HepG2-type
glucose transporters in rat kidney: immunofluorescence and immunogold study. J Histochem Cytochem 39: 287–298, 1991.
Taroni C, Jones S, Thornton JM. Analysis and prediction of
carbohydrate binding sites. Protein Eng 13: 89 –98, 2000.
Tarpey PS, Wood IS, Shirazi-Beechey SP, Beechey RB. Amino
acid sequence and the cellular location of the Na⫹-dependent
D-glucose symporters (SGLT1) in the ovine enterocyte and the
parotid acinar cell. Biochem J 312: 293–300, 1995.
Tasic V, Slaveska N, Blau N, Santer R. Nephrolithiasis in a child
with glucose-galactose malabsorption. Pediatr Nephrol 19: 244 –
246, 2004.
Tazawa S, Yamato T, Fujikura H, Hiratochi M, Itoh F, Tomae
M, Takemura Y, Maruyama H, Sugiyama T, Wakamatsu A,
Isogai T, Isaji M. SLC5A9/SGLT4, a new Na⫹-dependent glucose
transporter, is an essential transporter for mannose, 1,5-anhydroD-glucitol, and fructose. Life Sci 76: 1039 –1050, 2005.
Toggenburger G, Kessler M, Semenza G. Phlorizin as a probe of
the small-intestinal Na⫹,D-glucose cotransporter. A model. Biochim
Biophys Acta 688: 557–571, 1982.
Turk E, Gasymov OK, Lanza S, Horwitz J, Wright EM. A
reinvestigation of the secondary structure of functionally active
vSGLT, the Vibrio sodium/galactose cotransporter. Biochemistry
45: 1470 –1479, 2006.
Turk E, Kerner CJ, Lostao MP, Wright EM. Membrane topology
of the human Na⫹/glucose cotransporter SGLT1. J Biol Chem 271:
1925–1934, 1996.
Turk E, Kim O, le Coutre J, Whitelegge JP, Eskandari S, Lam
JT, Kreman M, Zampighi G, Faull KF, Wright EM. Molecular
characterization of Vibrio parahaemolyticus vSGLT: a model for
sodium-coupled sugar cotransporters. J Biol Chem 275: 25711–
25716, 2000.
Turk E, Klisak I, Bacallao R, Sparkes RS, Wright EM. Assignment of the human Na⫹/glucose cotransporter gene SGLT1 to
chromosome 22q13. 1 Genomics 17: 752–754, 1993.
Physiol Rev • VOL
793
220. Turk E, Martin MG, Wright EM. Structure of the human Na⫹/
glucose cotransporter gene SGLT1. J Biol Chem 269: 15204 –15209,
1994.
221. Turk E, Wright EM. Membrane topology motifs in the SGLT
cotransporter family. J Membr Biol 159: 1–20, 1997.
222. Turk E, Zabel B, Mundlos S, Dyer J, Wright EM. Glucose/
galactose malabsorption caused by a defect in the Na⫹/glucose
cotransporter. Nature 350: 354 –356, 1991.
223. Turner RJ, Moran A. Further studies of proximal tubular brush
border membrane D-glucose transport heterogeneity. J Membr Biol
70: 37– 45, 1982.
224. Turner RJ, Moran A. Heterogeneity of sodium-dependent D-glucose transport sites along the proximal tubule: evidence from
vesicle studies. Am J Physiol Renal Fluid Electrolyte Physiol 242:
F406 –F414, 1982.
225. Tyagi NK, Goyal P, Kumar A, Pandey D, Siess W, Kinne RK.
High-yield functional expression of human sodium/D-glucose
cotransporter1 in Pichia pastoris and characterization of ligandinduced conformational changes as studied by tryptophan fluorescence. Biochemistry 44: 15514 –15524, 2005.
226. Umbach JA, Coady MJ, Wright EM. Intestinal Na⫹/glucose
cotransporter expressed in Xenopus oocytes is electrogenic. Biophys J 57: 1217–1224, 1990.
227. Vallon V, Platt KA, Cunard R, Schroth J, Whaley J, Thomson
SC, Koepsell H, Rieg T. SGLT2 mediates glucose reabsorption in
the early proximal tubule. J Am Soc Nephrol 1–36, 2010.
228. Vayro S, Lo B, Silverman M. Functional studies of the rabbit
intestinal Na⫹/glucose carrier (SGLT1) expressed in COS-7 cells:
evaluation of the mutant A166C indicates this region is important
for Na⫹-activation of the carrier. Biochem J 332: 119 –125, 1998.
229. Vayro S, Wood IS, Dyer J, Shirazi-Beechey SP. Transcriptional
regulation of the ovine intestinal Na⫹/glucose cotransporter SGLT1
gene. Role of HNF-1 in glucose activation of promoter function.
Eur J Biochem 268: 5460 –5470, 2001.
230. Veenstra M, Lanza S, Hirayama BA, Turk E, Wright EM. Local
conformational changes in the Vibrio Na⫹/galactose cotransporter.
Biochemistry 43: 3620 –3627, 2004.
231. Veenstra M, Turk E, Wright EM. A ligand-dependent conformational change of the Na⫹/galactose cotransporter of Vibrio parahaemolyticus, monitored by tryptophan fluorescence. J Membr Biol
185: 249 –255, 2002.
232. Veyhl M, Keller T, Gorboulev V, Vernaleken A, Koepsell H.
RS1 (RSC1A1) regulates the exocytotic pathway of Na⫹-D-glucose
cotransporter SGLT1. Am J Physiol Renal Physiol 291: F1213–
F1223, 2006.
233. Veyhl M, Wagner CA, Gorboulev V, Schmitt BM, Lang F,
Koepsell H. Downregulation of the Na⫹-D-glucose cotransporter
SGLT1 by protein RS1 (RSC1A1) is dependent on dynamin and
protein kinase C. J Membr Biol 196: 71– 81, 2003.
234. Veyhl M, Wagner K, Volk C, Gorboulev V, Baumgarten K,
Weber WM, Schaper M, Bertram B, Wiessler M, Koepsell H.
Transport of the new chemotherapeutic agent beta-D-glucosylisophosphoramide mustard (D-19575) into tumor cells is mediated by
the Na⫹-D-glucose cotransporter SAAT1. Proc Natl Acad Sci USA
95: 2914 –2919, 1998.
235. Voss AA, Diez-Sampedro A, Hirayama BA, Loo DDF, Wright
EM. Imino sugars are potent agonists of the human glucose sensor
SGLT3. Mol Pharmacol 71: 628 – 634, 2007.
236. Wang J, Cortina G, Wu SV, Tran R, Cho JH, Tsai MJ, Bailey
TJ, Jamrich M, Ament ME, Treem WR, Hill ID, Vargas JH,
Gershman G, Farmer DG, Reyen L, Martin MG. Mutant neurogenin-3 in congenital malabsorptive diarrhea. N Engl J Med 355:
270 –280, 2006.
237. Washburn WN. Development of the renal glucose reabsorption
inhibitors: a new mechanism for the pharmacotherapy of diabetes
mellitus type 2. J Med Chem 52: 1785–1794, 2009.
238. Wanatabe A, Choe S, Chaptal V, Rosenberg JM, Wright EM,
Grabe M, Abramson J. The mechanism of sodium and substrate
release from the binding pocket of vSGLT. Nature 468: 988 –991,
2010.
239. Weihua Z, Tsan R, Huang WC, Wu Q, Chiu CH, Fidler IJ, Hung
MC. Survival of cancer cells is maintained by EGFR independent of
its kinase activity. Cancer Cell 13: 385–393, 2008.
91 • APRIL 2011 •
www.prv.org
794
WRIGHT, LOO, AND HIRAYAMA
240. Wells RG, Pajor AM, Kanai Y, Turk E, Wright EM, Hediger
MA. Cloning of a human kidney cDNA with similarity to the
sodium-glucose cotransporter. Am J Physiol Renal Fluid Electrolyte Physiol 263: F459 –F465, 1992.
241. Weyand S, Shimamura T, Yajima S, Suzuki S, Mirza O, Krusong K, Carpenter EP, Rutherford NG, Hadden JM, O’Reilly J,
Ma P, Saidijam M, Patching SG, Hope RJ, Norbertczak HT,
Roach PC, Iwata S, Henderson PJ, Cameron AD. Structure and
molecular mechanism of a nucleobase-cation-symport-1 family
transporter. Science 322: 709 –713, 2008.
242. Wielert-Badt S, Lin JT, Lorenz M, Fritz S, Kinne RK. Probing
the conformation of the sugar transport inhibitor phlorizin by
2D-NMR, molecular dynamics studies, and pharmacophore analysis. J Med Chem 43: 1692–1698, 2000.
243. Wood IS, Dyer J, Hofmann RR, Shirazi-Beechey SP. Expression of the Na⫹/glucose co-transporter (SGLT1) in the intestine of
domestic and wild ruminants. Pflügers Arch 441: 155–162, 2000.
244. Wright EM. Diseases of renal glucose handling. In: Genetic Diseases of the Kidney, edited by Lifton R. New York: Elsevier, 2008,
p. 1130 –1140.
245. Wright EM. Renal Na⫹-glucose cotransporters Am J Physiol Renal Physiol 280: F10 –F18, 2001.
246. Wright EM, Hirayama BA, Loo DDF. Active sugar transport in
health and disease. J Intern Med 261: 32– 43, 2007.
247. Wright EM, Hirayama BA, Loo DDF, Turk E, Hager K. Intestinal sugar transport. In: Physiology of Gastrointestinal Tract (3rd
ed.), edited by Johnson E. New York: Raven, 1994, p. 1751–1772.
248. Wright EM, Hirsch JR, Loo DDF, Zampighi GA. Regulation of
Na⫹/glucose cotransporters. J Exp Biol New York: 200: 287–293,
1997.
249. Wright EM, Loo DDF, Hirayama BA, Turk E. Surprising versatility of Na⫹-glucose cotransporters: SLC5. Physiology 19: 370 –376,
2004.
250. Wright EM, Martin MG, Turk E. Familial glucose-galactose malabsorption and hereditary renal glycosuria. In: The Metabolic and
Molecular Bases of Inherited Disease, edited by Scriver CR, Beaudet AL, Sly WS, Valle D, Childs B, Kinzler KW, Vogelstein B. New
York: McGraw-Hill, p. 4891– 4908.
251. Wright EM, Turk E. The sodium/glucose cotransport family SLC5.
Pflügers Arch 447: 510 –518, 2004.
252. Wright EM, van Os CH, Mircheff AK. Sugar uptake by intestinal
basolateral membrane vesicles. Biochim Biophys Acta 597: 112–
124, 1980.
253. Wright EM, Turk E. The sodium glucose cotransport family SLC5.
Pflügers Arch 510 –518, 2004.
254. Xie Z, Turk E, Wright EM. Characterization of the Vibrio parahaemolyticus Na⫹/glucose cotransporter. A bacterial member of
Physiol Rev • VOL
255.
256.
257.
258.
259.
260.
261.
262.
263.
264.
265.
266.
the sodium/glucose transporter (SGLT) family. J Biol Chem 275:
25959 –25964, 2000.
Xu B, Lv B, Feng Y, Xu G, Du J, Welihinda A, Sheng Z, Seed B,
Chen Y. O-Spiro C-aryl glucosides as novel sodium-dependent
glucose co-transporter 2 (SGLT2) inhibitors. Bioorg Med Chem Lett
19: 5632–5635, 2009.
Yamashita A, Singh SK, Kawate T, Jin Y, Gouaux E. Crystal
structure of a bacterial homologue of Na⫹/Cl⫺-dependent neurotransmitter transporters. Nature 437: 215–223, 2005.
You G, Lee WS, Barros EJ, Kanai Y, Huo TL, Khawaja S, Wells
RG, Nigam SK, Hediger MA. Molecular characteristics of Na⫹coupled glucose transporters in adult and embryonic rat kidney. J
Biol Chem 270: 29365–29371, 1995.
Yu AS, Hirayama BA, Timbol G, Liu J, Basarah E, Kepe V,
Satyamurthy N, Wright EM, Huang SC, Barrio JR. Functional
expression of SGLTs in rat brain. Am J Physiol Cell Physiol 299:
C1277–C1284, 2010.
Yu LC, Flynn AN, Turner JR, Buret AG. SGLT-1-mediated glucose uptake protects intestinal epithelial cells against LPS-induced
apoptosis and barrier defects: a novel cellular rescue mechanism?
FASEB J 19: 1822–1835, 2005.
Yu LC, Huang CY, Kuo WT, Sayer H, Turner JR, Buret AG.
SGLT-1-mediated glucose uptake protects human intestinal epithelial cells against Giardia duodenalis-induced apoptosis. Int J
Parasitol 38: 923–934, 2008.
Zagotta WN, Hoshi T, Aldrich RW. Shaker potassium channel
gating. III. Evaluation of kinetic models for activation. J Gen
Physiol 103: 321–362, 1994.
Zampighi GA, Kreman M, Boorer KJ, Loo DDF, Bezanilla F,
Chandy G, Hall JE, Wright EM. A method for determining the
unitary functional capacity of cloned channels and transporters
expressed in Xenopus laevis oocytes. J Membr Biol 148: 65–78,
1995.
Zeuthen T. Water-transporting proteins. J Membr Biol 234: 57–73,
2010.
Zeuthen T, Belhage B, Zeuthen E. Water transport by Na⫹coupled cotransporters of glucose (SGLT1) and of iodide (NIS).
The dependence of substrate size studied at high resolution. J
Physiol 570: 485– 499, 2006.
Zeuthen T, Zeuthen E. The mechanism of water transport in
Na⫹-coupled glucose transporters expressed in Xenopus oocytes.
Biophys J 93: 1413–1419, 2007.
Zhao FQ, Keating AF. Expression and regulation of glucose
transporters in the bovine mammary gland. J Dairy Sci 90 Suppl 1:
E76 –E86, 2007.
91 • APRIL 2011 •
www.prv.org
Download